One of my F9 machines -- actually my #1 main machine -- started having display problems. I see this a lot, because my hardware isn't really up to date enough to handle my monitor optimally. So I tried a couple of tricks that usually help, and they didn't.
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
I can still do what I'm doing now: access the list via Pan on another machine, and go back & forth with my KVM switch; but it makes it hard, for instance, to post things like a list of contents for /var/log, or any whole file in it.
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
On Sun, 2008-10-26 at 20:12 +0000, Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems. I see this a lot, because my hardware isn't really up to date enough to handle my monitor optimally. So I tried a couple of tricks that usually help, and they didn't.
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
I can still do what I'm doing now: access the list via Pan on another machine, and go back & forth with my KVM switch; but it makes it hard, for instance, to post things like a list of contents for /var/log, or any whole file in it.
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
---- from command line, cut down your large log files by locating them...
# find /var/log -type f -size +2000k -exec ls -lh {} ; | awk '{ print $8 ": " $5 }'
and any file that is really large (i.e. /var/log/Xorg.0.log)
command line empty it... # > /var/log/Xorg.0.log
if startx only works as root, it sounds as if some permission isn't correct (possibly /etc/X11/xorg.conf is not 644, readable by all users). Of course if that were the issue, /var/log/Xorg.0.log would tell you that or if it were some other permissions issue, the problem would probably be listed in /var/log/messages
On Fedora 9, it often is enough to just do something like # mv /etc/X11/xorg.conf /etc/X11/xorg.conf-bak # shutdown now -r
and it is automatically rebuilt upon reboot
Craig
On Sun, 26 Oct 2008 13:54:54 -0700, Craig White wrote:
On Sun, 2008-10-26 at 20:12 +0000, Beartooth wrote:
[....]
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
from command line, cut down your large log files by locating them...
# find /var/log -type f -size +2000k -exec ls -lh {} ; | awk '{ print $8 ": " $5 }'
I can't see for sure how many spaces, if any, there are between "print" and "$8"; but I can experiment.
Meanwhile, however, I'm getting other troubles. One error message says "invalid argument + to size".
The man page is way beyond me to slog through in less than about six months; but I did find a + on it; so I tried using the other + key. No joy.
I also get an error message saying "missing argument to -exec"
and any file that is really large (i.e. /var/log/Xorg.0.log)
command line empty it... # > /var/log/Xorg.0.log
if startx only works as root, it sounds as if some permission isn't correct (possibly /etc/X11/xorg.conf is not 644, readable by all users). Of course if that were the issue, /var/log/Xorg.0.log would tell you that or if it were some other permissions issue, the problem would probably be listed in /var/log/messages
I'm very weak on permissions; I can never remember the mnemonic for those numbers. but I'll try "cd /etc/X11" and then "ls -l" to see if that tells me anything either comprehensible or repeatable.
On Fedora 9, it often is enough to just do something like # mv /etc/X11/xorg.conf /etc/X11/xorg.conf-bak # shutdown now -r
and it is automatically rebuilt upon reboot
I tried that, and got only the dread "out of range" message. I'll try again with "system-config-display" and "system-config-display -- reconfig" It would be a great help to get X back for my user, if only to enable c&p between the CLI and Pan.
Many thanks to all for the help so far!
On Mon, 2008-10-27 at 14:24 +0000, Beartooth wrote:
On Sun, 26 Oct 2008 13:54:54 -0700, Craig White wrote:
On Sun, 2008-10-26 at 20:12 +0000, Beartooth wrote:
[....]
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
from command line, cut down your large log files by locating them...
# find /var/log -type f -size +2000k -exec ls -lh {} ; | awk '{ print $8 ": " $5 }'
I can't see for sure how many spaces, if any, there are between "print" and "$8"; but I can experiment.
Meanwhile, however, I'm getting other troubles. One error message says "invalid argument + to size".
The man page is way beyond me to slog through in less than about six months; but I did find a + on it; so I tried using the other + key. No joy.
I also get an error message saying "missing argument to -exec"
---- it's just a simple command to print a list of all log files greater than 20 Mb.
If you can't manage to just type it out as it was given, just do 'ls -lh /var/log' and you can inspect each file for size. ----
and any file that is really large (i.e. /var/log/Xorg.0.log)
command line empty it... # > /var/log/Xorg.0.log
if startx only works as root, it sounds as if some permission isn't correct (possibly /etc/X11/xorg.conf is not 644, readable by all users). Of course if that were the issue, /var/log/Xorg.0.log would tell you that or if it were some other permissions issue, the problem would probably be listed in /var/log/messages
I'm very weak on permissions; I can never remember the mnemonic for those numbers. but I'll try "cd /etc/X11" and then "ls -l" to see if that tells me anything either comprehensible or repeatable.
---- read = 4 write = 2 exec = 1
rwxrwxrwx ^^^^^^^^^ | | | | | other | group owner
thus rwx______ is 700 rw_rw_rw_ is 444 rw_r__r__ is 644 ----
On Fedora 9, it often is enough to just do something like # mv /etc/X11/xorg.conf /etc/X11/xorg.conf-bak # shutdown now -r
and it is automatically rebuilt upon reboot
I tried that, and got only the dread "out of range" message. I'll try again with "system-config-display" and "system-config-display -- reconfig" It would be a great help to get X back for my user, if only to enable c&p between the CLI and Pan.
Many thanks to all for the help so far!
---- seems odd because reports I have heard about Fedora 9 is that it automatically builds xorg.conf if not present.
init 3 log in as root system-config-display --reconfig init 5
should do the trick though
Craig
On Mon, 2008-10-27 at 07:39 -0700, Craig White wrote:
seems odd because reports I have heard about Fedora 9 is that it automatically builds xorg.conf if not present.
The reports are wrong or you misunderstood them. X11 can start with no xorg.conf file (it's often the best way) but it doesn't create one if it's not there, it just configures itself internally every time it starts up.
poc
On Mon, 2008-10-27 at 07:39 -0700, Craig White wrote:
read = 4 write = 2 exec = 1
rwxrwxrwx ^^^^^^^^^ | | | | | other | group owner
thus rwx______ is 700
rw_rw_rw_ is 444
Above should be 666?
rw_r__r__ is 644
On Mon, 2008-10-27 at 09:52 -0500, Mike Chambers wrote:
On Mon, 2008-10-27 at 07:39 -0700, Craig White wrote:
read = 4 write = 2 exec = 1
rwxrwxrwx ^^^^^^^^^ | | | | | other | group owner
thus rwx______ is 700
rw_rw_rw_ is 444Above should be 666?
rw_r__r__ is 644
---- thanks - my bad
Craig
On Mon, 27 Oct 2008 07:39:01 -0700, Craig White wrote:
from command line, cut down your large log files by locating them...
# find /var/log -type f -size +2000k -exec ls -lh {} ; | awk '{ print $8 ": " $5 }'
[...]
it's just a simple command to print a list of all log files greater than 20 Mb.
If you can't manage to just type it out as it was given, just do 'ls -lh /var/log' and you can inspect each file for size. ----
As you'll've seen by now, I found a typo and got it.
and any file that is really large (i.e. /var/log/Xorg.0.log)
command line empty it... # > /var/log/Xorg.0.log
if startx only works as root, it sounds as if some permission isn't correct (possibly /etc/X11/xorg.conf is not 644, readable by all users). Of course if that were the issue, /var/log/Xorg.0.log would tell you that or if it were some other permissions issue, the problem would probably be listed in /var/log/messages
I'm very weak on permissions; I can never remember the mnemonic for those numbers. but I'll try "cd /etc/X11" and then "ls -l" to see if that tells me anything either comprehensible or repeatable.
read = 4 write = 2 exec = 1
rwxrwxrwx ^^^^^^^^^ | | | | | other | group owner
thus rwx______ is 700 rw_rw_rw_ is 444 rw_r__r__ is 644
I've re-checked, and I think my monitor newer than any computer is biting me again. xorg.conf seems not to exist; xorg.conf-bak is -rw-r-- r-- 1
On Fedora 9, it often is enough to just do something like # mv /etc/X11/xorg.conf /etc/X11/xorg.conf-bak # shutdown now -r
and it is automatically rebuilt upon reboot
I tried that, and got only the dread "out of range" message. I'll try again with "system-config-display" and "system-config-display -- reconfig" It would be a great help to get X back for my user, if only to enable c&p between the CLI and Pan.
seems odd because reports I have heard about Fedora 9 is that it automatically builds xorg.conf if not present.
init 3 log in as root system-config-display --reconfig init 5
should do the trick though
The problem is that wide monitors had not yet come on the market when my newest machine was built. I have an hp w2207h, which is 1680x1050; I can run it usable by setting the hardware to lcd that size, and then the display to 1280x1024; that lets the monitor stretch it sidewise. But both system-config-display and system-config-display -- reconfig merely stymie things; I get the out of range message and have to hit Ctrl-Alt-Backspace, or else the monitor goes black and stays black.
I haven't tried init commands lately; but grub may be set to init 3; I've done that before, in order at least to get a prompt.
I also took a quick look from the CLI at that messages file, and couldn't spot anything. I'll go do startx as root and try to get a display that lets me skim enough of them to spot a pattern.
On Mon, 2008-10-27 at 15:06 +0000, Beartooth wrote:
On Mon, 27 Oct 2008 07:39:01 -0700, Craig White wrote:
init 3 log in as root system-config-display --reconfig init 5
should do the trick though
The problem is that wide monitors had not yet come on the market when my newest machine was built. I have an hp w2207h, which is 1680x1050; I can run it usable by setting the hardware to lcd that size, and then the display to 1280x1024; that lets the monitor stretch it sidewise. But both system-config-display and system-config-display -- reconfig merely stymie things; I get the out of range message and have to hit Ctrl-Alt-Backspace, or else the monitor goes black and stays black.
---- you really need to get your video card configured properly to use a 1680x1050 wide panel display. I'm not certain that the xorg supplied nv/radeon drivers can do that and it may take the binary blob drivers from nVidia or ATI ----
I haven't tried init commands lately; but grub may be set to init 3; I've done that before, in order at least to get a prompt.
I also took a quick look from the CLI at that messages file, and couldn't spot anything. I'll go do startx as root and try to get a display that lets me skim enough of them to spot a pattern.
---- viewing big log files in GUI is slower than molassess in the middle of January.
Starting at the bottom of the logs, errors are being continuously being repeated making the logs grow like crazy.
Craig
On Sun, 26 Oct 2008 13:54:54 -0700, Craig White wrote:
[...]
What should I do first??
from command line, cut down your large log files by locating them...
# find /var/log -type f -size +2000k -exec ls -lh {} ; | awk '{ print $8 ": " $5 }'
Proofreading for the third time, I discovered a missing space; now I get no error messages, and it tells me the culprit is /var/log/ messages for yesterday -- with 54GB all by itself. The poor thing must've been going bonkers; gawdknows why ...
and any file that is really large (i.e. /var/log/Xorg.0.log)
command line empty it... # > /var/log/Xorg.0.log
Is there some way to find out what makes that message file so big -- i.e., why the machine was losing its mind? I can't hope to read 54 gigs in my lifetime ...
On Mon, 2008-10-27 at 14:32 +0000, Beartooth wrote:
Is there some way to find out what makes that message file so big -- i.e., why the machine was losing its mind? I can't hope to read 54 gigs in my lifetime ...
"tail -n 1000 /var/log/messages" will show the last 1000 lines of the log. If one message is repeating over and over (as Chris pointed out), it should be pretty obvious.
Jonathan
On Mon, 2008-10-27 at 14:32 +0000, Beartooth wrote:
Is there some way to find out what makes that message file so big -- i.e., why the machine was losing its mind? I can't hope to read 54 gigs in my lifetime ...
---- less /var/log/messages
press '>' to get to end of messages use 'w' or 'Page Up' to page up
press 'Control-C' to stop line number calculations (pointless in 54G file)
read from the bottom, it should be obvious what the problem is
Craig
On Mon, 27 Oct 2008 07:41:26 -0700, Craig White wrote:
On Mon, 2008-10-27 at 14:32 +0000, Beartooth wrote:
Is there some way to find out what makes that message file so big -- i.e., why the machine was losing its mind? I can't hope to read 54 gigs in my lifetime ...
less /var/log/messages
press '>' to get to end of messages use 'w' or 'Page Up' to page up
press 'Control-C' to stop line number calculations (pointless in 54G file)
read from the bottom, it should be obvious what the problem is
I may have messed up. When suddenly startx failed even for root, I tried "> /var/log/messages" (I think it was) and now the command to look for the space-grabber returns nothing near so large. But then everything else I tried kept hitting the out of range display. So I finally hit the reset button.
After rebooting that way, it now logs me straight in, GUI and all -- but not, I think, with optimal display.
All I recall from the bloated stuff was lots of differing and interspersed lines about USB devices, and one of them not doing fax. (I never use fax.)
Is there a good way to keep an eye on /var/log/messages today, and catch it in time if it happens again?
On Mon, 2008-10-27 at 15:18 +0000, Beartooth wrote:
Is there a good way to keep an eye on /var/log/messagestoday, and catch it in time if it happens again?
Until this is resolved, I would recommend running "tail -f /var/log/messages" in a console session or xterm, and taking a look at it now and again.
poc
On Mon, 27 Oct 2008 12:14:37 -0430, Patrick O'Callaghan wrote:
On Mon, 2008-10-27 at 15:18 +0000, Beartooth wrote:
Is there a good way to keep an eye on /var/log/messagestoday, and catch it in time if it happens again?
Until this is resolved, I would recommend running "tail -f /var/log/messages" in a console session or xterm, and taking a look at it now and again.
Craig White had recommended :
# > /var/log/messages that will completely empty it out
and I did. Starting after that, I get this, so far (without, nota bene, having laid a finger on any usb anything, much less tried to fax anything anywhere) :
===== ===== ===== ===== [root@Hbsk2 ~]# tail -f /var/log/messages Oct 27 13:04:42 Hbsk2 kernel: usb 2-1: USB disconnect, address 6 Oct 27 13:04:42 Hbsk2 kernel: usb 2-1.2: USB disconnect, address 7 Oct 27 13:04:42 Hbsk2 kernel: usblp0: removed Oct 27 13:04:42 Hbsk2 kernel: usb 2-1.3: USB disconnect, address 8 Oct 27 13:04:42 Hbsk2 kernel: usb 2-1.4: USB disconnect, address 9 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1: new full speed USB device using ohci_hcd and address 10 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1: configuration #1 chosen from 1 choice Oct 27 13:07:27 Hbsk2 kernel: hub 2-1:1.0: USB hub found Oct 27 13:07:27 Hbsk2 kernel: hub 2-1:1.0: 4 ports detected Oct 27 13:07:27 Hbsk2 kernel: usb 2-1: New USB device found, idVendor=0451, idProduct=2046 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: new full speed USB device using ohci_hcd and address 11 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: configuration #1 chosen from 1 choice Oct 27 13:07:27 Hbsk2 kernel: usblp0: USB Bidirectional printer dev 11 if 1 alt 0 proto 2 vid 0x03F0 pid 0x3F11 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: New USB device found, idVendor=03f0, idProduct=3f11 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: New USB device strings: Mfr=1, Product=2, SerialNumber=3 Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: Product: psc 1310 series Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: Manufacturer: hp Oct 27 13:07:27 Hbsk2 kernel: usb 2-1.2: SerialNumber: CN4AIC71YKO2 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: new low speed USB device using ohci_hcd and address 12 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: configuration #1 chosen from 1 choice Oct 27 13:07:28 Hbsk2 kernel: input: HP Mouse HP Mouse as /devices/ pci0000:00/0000:00:02.0/usb2/2-1/2-1.3/2-1.3:1.0/input/input10 Oct 27 13:07:28 Hbsk2 kernel: input,hidraw0: USB HID v1.00 Mouse [HP Mouse HP Mouse] on usb-0000:00:02.0-1.3 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: New USB device found, idVendor=03f0, idProduct=0c1d Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: New USB device strings: Mfr=1, Product=1, SerialNumber=0 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: Product: HP Mouse Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.3: Manufacturer: HP Mouse Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: new low speed USB device using ohci_hcd and address 13 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: configuration #1 chosen from 1 choice Oct 27 13:07:28 Hbsk2 kernel: input: BTC USB Multimedia Keyboard as / devices/pci0000:00/0000:00:02.0/usb2/2-1/2-1.4/2-1.4:1.0/input/input11 Oct 27 13:07:28 Hbsk2 kernel: input,hidraw1: USB HID v1.10 Keyboard [BTC USB Multimedia Keyboard] on usb-0000:00:02.0-1.4 Oct 27 13:07:28 Hbsk2 kernel: input: BTC USB Multimedia Keyboard as / devices/pci0000:00/0000:00:02.0/usb2/2-1/2-1.4/2-1.4:1.1/input/input12 Oct 27 13:07:28 Hbsk2 kernel: input,hiddev96,hidraw2: USB HID v1.10 Device [BTC USB Multimedia Keyboard] on usb-0000:00:02.0-1.4 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: New USB device found, idVendor=046d, idProduct=c313 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: New USB device strings: Mfr=1, Product=2, SerialNumber=0 Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: Product: USB Multimedia Keyboard Oct 27 13:07:28 Hbsk2 kernel: usb 2-1.4: Manufacturer: BTC Oct 27 13:07:31 Hbsk2 python: hp-makeuri[5232]: error: Device does not support fax. Oct 27 13:07:31 Hbsk2 python: hp-makeuri[5234]: error: Device does not support fax. ===== ===== ===== =====
It sure looks to me like something is going on that shouldn't be.
On Mon, 2008-10-27 at 17:40 +0000, Beartooth wrote:
and I did. Starting after that, I get this, so far (without, nota bene, having laid a finger on any usb anything, much less tried to fax anything anywhere) :
You sure you don't have one of those All in One type printers or something attached via usb to your machine?
On Mon, 27 Oct 2008 14:08:15 -0500, Mike Chambers wrote:
On Mon, 2008-10-27 at 17:40 +0000, Beartooth wrote:
and I did. Starting after that, I get this, so far (without, nota bene, having laid a finger on any usb anything, much less tried to fax anything anywhere) :
You sure you don't have one of those All in One type printers or something attached via usb to your machine?
Thanks for the cue! I was just going to post about that.
First, I decided to run tail -f /var/log/messages on all four main machines (i.e., those behind my KVM switch), on the theory that keeping an eye on all would make any anomaly more likely to stand out.
What I'm finding is, on the contrary, that the messages look very similar on all four.
Now as to hardware -- which has been in place for many months with few or no problems.
The KVM switch is a MiniView G-CSIO4U, a 4-port USB switch with two extra USB ports in back, into which the manufacturer says you can connect any two peripherals you want to share among your machines.
The printer (an HP PSC 1315v, also USB)) has been in one of those ports for an age. My external DVD RW drive gets plugged into and out of the other port all the time; but the printer stays on the same connection. (The keyboard, mouse (which is indeed an hp), and monitor all get shifted to one individual machine when I do an install or upgrade (so that the monitor and PC can negotiate settings), and otherwise stay on the switch.
Marginal note : I've noticed that, if I hit the KVM switch's button while one PC is printing, the printer stops till I come back to that machine.
Absent-mindedness is the occupational hazard of the professoriate, and like so much else is exacerbated with age; so it's possible that I was doing something odd yesterday and don't remember -- or, of course, didn't realize it was odd. But fwiw, I think not.
Marginal note 2 : it is quite common, certainly with me and I believe with others, for everything to hang for some seconds every time I use the KVM switch; and the mouse is usually the last thing I see recognized. (I print very little, and have no idea how long it takes to recognize the printer.)
Finally, an hour or so ago, I tried turning the printer off with its power switch. Since that time, the messages have become fewer, but not stopped.
Beartooth wrote:
The KVM switch is a MiniView G-CSIO4U, a 4-port USB switch with two extra USB ports in back, into which the manufacturer says you can connect any two peripherals you want to share among your machines.
The printer (an HP PSC 1315v, also USB)) has been in one of those ports for an age. My external DVD RW drive gets plugged into and out of the other port all the time; but the printer stays on the same connection. (The keyboard, mouse (which is indeed an hp), and monitor all get shifted to one individual machine when I do an install or upgrade (so that the monitor and PC can negotiate settings), and otherwise stay on the switch.
Switching the KVM switch is equivalent to unplugging all the devices from one computer and plugging them into another. Linux will print some messages every time you plug in or remove a USB device, but you'd have to be switching like crazy to produce 54 GB of messages that way. I suppose a loose cable might make it seem like all the devices are constantly plugged in and removed, but I still don't quite see how the log could grow that big. The excerpt you posted was 3411 characters. Printing all of that once a second for a week would still produce only two gigabytes.
Marginal note : I've noticed that, if I hit the KVM switch's button while one PC is printing, the printer stops till I come back to that machine.
Of course. Cups has to stop printing when you essentially unplug the printer, and then it continues when you plug the printer in again.
Finally, an hour or so ago, I tried turning the printer off with its power switch. Since that time, the messages have become fewer, but not stopped.
I'd imagine that the messages about the printer have stopped, and the ones about the keyboard, the mouse and the hub continue. (There's a USB hub inside the KVM switch.) Every time you switch to another machine to look for new messages, you cause more messages.
Those messages aren't errors and you don't need to worry about them as long as the log doesn't grow out of control again. It's quite possible that most of those 54 GB was something completely different that hasn't resurfaced yet. I'd recommend doing "ls -l /var/log/messages*" now and then to keep an eye on it, and investigate further if it grows to many megabytes.
Björn Persson
Björn Persson wrote:
I'd imagine that the messages about the printer have stopped, and the ones about the keyboard, the mouse and the hub continue. (There's a USB hub inside the KVM switch.) Every time you switch to another machine to look for new messages, you cause more messages.
Those messages aren't errors and you don't need to worry about them as long as the log doesn't grow out of control again. It's quite possible that most of those 54 GB was something completely different that hasn't resurfaced yet. I'd recommend doing "ls -l /var/log/messages*" now and then to keep an eye on it, and investigate further if it grows to many megabytes.
Björn Persson
I have not been really been following this thread, but I do remember something about having to go and read root's messages. I got the impression that root's mail never gets read. If all root's mail is building up in his mail box, and it never gets cleaned out, it is probably a significant cause of the problem as well. (I wounder how many "running out of disk space" messages root has...) Part of the fix should probably include setting up an alias so root's mail go to a normal user. If it is not going to be read at all, then maybe forward it to /dev/null. :)
Mikkel
On Mon, 27 Oct 2008 20:15:06 -0500, Mikkel L. Ellertson wrote:
Björn Persson wrote:
I'd imagine that the messages about the printer have stopped, and the ones about the keyboard, the mouse and the hub continue. (There's a USB hub inside the KVM switch.) Every time you switch to another machine to look for new messages, you cause more messages.
Those messages aren't errors and you don't need to worry about them as long as the log doesn't grow out of control again. It's quite possible that most of those 54 GB was something completely different that hasn't resurfaced yet. I'd recommend doing "ls -l /var/log/messages*" now and then to keep an eye on it, and investigate further if it grows to many megabytes.
Björn Persson
I have not been really been following this thread, but I do remember something about having to go and read root's messages. I got the impression that root's mail never gets read. If all root's mail is building up in his mail box, and it never gets cleaned out, it is probably a significant cause of the problem as well. (I wounder how many "running out of disk space" messages root has...) Part of the fix should probably include setting up an alias so root's mail go to a normal user. If it is not going to be read at all, then maybe forward it to /dev/null. :)
Well, it's true, alas!, that I hardly ever think to read root's mail; otoh, wouldn't it show up in baobab if root's mail were getting bloated? That's how the bloat in /var/log/messages showed up.
I'm online too much as it is; and most logs are nearly gibberish to me.
Beartooth wrote:
On Mon, 27 Oct 2008 20:15:06 -0500, Mikkel L. Ellertson wrote:
I have not been really been following this thread, but I do remember something about having to go and read root's messages. I got the impression that root's mail never gets read. If all root's mail is building up in his mail box, and it never gets cleaned out, it is probably a significant cause of the problem as well. (I wounder how many "running out of disk space" messages root has...) Part of the fix should probably include setting up an alias so root's mail go to a normal user. If it is not going to be read at all, then maybe forward it to /dev/null. :)
Well, it's true, alas!, that I hardly ever think to read root's mail; otoh, wouldn't it show up in baobab if root's mail were getting bloated? That's how the bloat in /var/log/messages showed up.
I'm online too much as it is; and most logs are nearly gibberish to me.
Well, the logwatch messages are usually fairly clear. For example, here is the disk usage report for this machine:
--------------------- Disk Space Begin ------------------------
Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 48G 8.5G 37G 19% / /dev/mapper/VolGroup00-LogVol01 141G 25G 109G 19% /home /dev/sda3 82G 55G 23G 71% /shared /dev/sda1 99M 19M 75M 21% /boot /shared/Fedora-8.iso 3.3G 3.3G 0 100% /var/www/html/Fedora8 /shared/Fedora-9.iso 3.4G 3.4G 0 100% /var/www/html/Fedora9
/shared/Fedora-8.iso => 100% Used. Warning. Disk Filling up. /shared/Fedora-9.iso => 100% Used. Warning. Disk Filling up.
---------------------- Disk Space End -------------------------
As you can see, it is complaining about the two loop-mounted .iso images are out of free space. :) The rest of the sections are about this clear. It is worth reading just for advanced working of problems, or strange things going on.
Mikkel
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Björn Persson wrote: <snip>
Switching the KVM switch is equivalent to unplugging all the devices from one computer and plugging them into another. Linux will print some messages every time you plug in or remove a USB device
as a 'side note', this should serve as a warning to anyone considering using a kvm switch.
non usb kvm switches do not have this problem. i have a ps/2 kvm switch that works great and yet to have any problems when switching between systems. - --
tc,hago.
g .
in a free world without fences, who needs gates.
learn linux: 'Rute User's Tutorial and Exposition' http://rute.2038bug.com/index.html.gz 'The Linux Documentation Project' http://www.tldp.org/ 'HowtoForge' http://howtoforge.com/
On Tue, 2008-10-28 at 16:48 +0000, g wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Björn Persson wrote:
<snip> > Switching the KVM switch is equivalent to unplugging all the devices from one > computer and plugging them into another. Linux will print some messages every > time you plug in or remove a USB device
---- no - a kvm should provide a continuous connection to each system but only route mouse/keyboard events when the proper terminal is selected. Other USB device event routing is sort of up to the imagination of the device developers as there really isn't much of a standard here.
I clearly don't have problems switching around systems with KVM including USB and/or PS/2 keyboard & mice & even a Wacom tablet ----
as a 'side note', this should serve as a warning to anyone considering using a kvm switch.
non usb kvm switches do not have this problem. i have a ps/2 kvm switch that works great and yet to have any problems when switching between systems.
---- it would help if you followed the conversation...the problem is not usb kvm which are actually becoming the norm because most modern motherboards do not even bother with ps/2 connectors, but rather that his particular kvm offers the ability to become a device hub which theoretically, meant that you could plug devices such as usb printers and they were for most purposes, shared. The devil of course is in the details and apparently the details aren't handled too well on that particular kvm.
Craig
Craig White wrote:
the problem is not usb kvm which are actually becoming the norm because most modern motherboards do not even bother with ps/2 connectors, but rather that his particular kvm offers the ability to become a device hub which theoretically, meant that you could plug devices such as usb printers and they were for most purposes, shared. The devil of course is in the details and apparently the details aren't handled too well on that particular kvm.
I'm not convinced that there is any problem with the KVM switch at all. USB is designed for hot plugging, so operating systems have to expect that a USB device may disappear at any time, and apparently Cups handles the disappering printer quite well. The USB related messages aren't filling the disk now, so why assume that they did before?
Unfortunately Beartooth had to nuke the huge file to get a usable system back, so if the problem doesn't resurface we may never know what it was.
Björn Persson
On Tue, 2008-10-28 at 23:10 +0100, Björn Persson wrote:
Craig White wrote:
the problem is not usb kvm which are actually becoming the norm because most modern motherboards do not even bother with ps/2 connectors, but rather that his particular kvm offers the ability to become a device hub which theoretically, meant that you could plug devices such as usb printers and they were for most purposes, shared. The devil of course is in the details and apparently the details aren't handled too well on that particular kvm.
I'm not convinced that there is any problem with the KVM switch at all. USB is designed for hot plugging, so operating systems have to expect that a USB device may disappear at any time, and apparently Cups handles the disappering printer quite well. The USB related messages aren't filling the disk now, so why assume that they did before?
---- it really makes no sense to be plugging printers in/out by KVM switch when it's so much more logical to connect it to one and share it, if for no other reason that it means that you only have to set it up on one system and once it's shared, the other computers find it and can use it automatically. ----
Unfortunately Beartooth had to nuke the huge file to get a usable system back, so if the problem doesn't resurface we may never know what it was.
---- seems to me that the solution may very well have been the problem though it may not have been the USB printer device, it might have been the multi-function options (i.e. Fax) which is what led us to asking about the device itself. It would have been nice had OP shown just a little skill to read through system logs to help troubleshoot but c'est la vie.
Craig
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Craig White wrote:
On Tue, 2008-10-28 at 16:48 +0000, g wrote:
Switching the KVM switch is equivalent to unplugging all the devices from one computer and plugging them into another. Linux will print some messages every time you plug in or remove a USB device
no - a kvm should provide a continuous connection to each system but only route mouse/keyboard events when the proper terminal is selected. Other USB device event routing is sort of up to the imagination of the device developers as there really isn't much of a standard here.
I clearly don't have problems switching around systems with KVM including USB and/or PS/2 keyboard & mice & even a Wacom tablet
as a 'side note', this should serve as a warning to anyone considering using a kvm switch.
non usb kvm switches do not have this problem. i have a ps/2 kvm switch that works great and yet to have any problems when switching between systems.
it would help if you followed the conversation...the problem is not usb
excuse me craig. i had not noted that this was 28th day of month. did you run out of mydol again?
if you look at above again, you will note *as a 'side note'*. which was in following of conversation, where in which you inserted a reply. it did not follow with 'subject:'. - --
tc,hago.
g .
in a free world without fences, who needs gates.
learn linux: 'Rute User's Tutorial and Exposition' http://rute.2038bug.com/index.html.gz 'The Linux Documentation Project' http://www.tldp.org/ 'HowtoForge' http://howtoforge.com/
On Wed, 2008-10-29 at 04:40 +0000, g wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Craig White wrote:
On Tue, 2008-10-28 at 16:48 +0000, g wrote:
Switching the KVM switch is equivalent to unplugging all the devices from one computer and plugging them into another. Linux will print some messages every time you plug in or remove a USB device
no - a kvm should provide a continuous connection to each system but only route mouse/keyboard events when the proper terminal is selected. Other USB device event routing is sort of up to the imagination of the device developers as there really isn't much of a standard here.
I clearly don't have problems switching around systems with KVM including USB and/or PS/2 keyboard & mice & even a Wacom tablet
as a 'side note', this should serve as a warning to anyone considering using a kvm switch.
non usb kvm switches do not have this problem. i have a ps/2 kvm switch that works great and yet to have any problems when switching between systems.
it would help if you followed the conversation...the problem is not usb
excuse me craig. i had not noted that this was 28th day of month. did you run out of mydol again?
if you look at above again, you will note *as a 'side note'*. which was in following of conversation, where in which you inserted a reply. it did not follow with 'subject:'.
---- so that means that if you preface comments as side notes, that they can ignore the reality that most computers/motherboards do not come with PS/2 connections any more and just offer USB for mouse/keyboard connections?
If you buy a KVM that doesn't service USB mouse/keyboard connections, it will be obsolete very soon - rendering your warning somewhat useless - that was my point - sorry you missed it.
Craig
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Craig White wrote: <snip>
so that means that if you preface comments as side notes,
still have not found your mydol?
If you buy a KVM that doesn't service USB mouse/keyboard connections,
being that systems i am using are non usb k/m i have no need for one with usb ability. when i do have multi systems using usb k/m, then i will buy proper kvm switch.
contrary to your thinking, rest of world of is not inversely proportioned to your ways and abilities. - --
tc,hago.
g .
in a free world without fences, who needs gates.
learn linux: 'Rute User's Tutorial and Exposition' http://rute.2038bug.com/index.html.gz 'The Linux Documentation Project' http://www.tldp.org/ 'LDP HOWTO-index' http://www.tldp.org/HOWTO/HOWTO-INDEX/index.html 'HowtoForge' http://howtoforge.com/
On Thu, 2008-10-30 at 14:30 +0000, g wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Craig White wrote:
<snip> > so that means that if you preface comments as side notes,
still have not found your mydol?
---- wow, add a sexist slur to an insult - classy ----
If you buy a KVM that doesn't service USB mouse/keyboard connections,
being that systems i am using are non usb k/m i have no need for one with usb ability. when i do have multi systems using usb k/m, then i will buy proper kvm switch.
contrary to your thinking, rest of world of is not inversely proportioned to your ways and abilities.
---- perhaps but if they have purchased computers in the last 2 years, they probably are already well tuned in to the fact that PS/2 keyboards and mice on soon to be obsolete...thus rendering your admonition that people not buy a KVM with USB as being likewise obsolete.
I'm not gonna bother researching but the last few PowerEdge servers that I bought from Dell were USB only...no PS/2. Likewise Optiplex systems from Dell have been USB only for quite some time. It's a good thing that you weren't part of the team trip planning for Christopher Columbus.
Craig
On Tue, 28 Oct 2008 00:54:22 +0100, Björn Persson wrote:
I Beartooth wrote:
[....]
Switching the KVM switch is equivalent to unplugging all the devices from one computer and plugging them into another. Linux will print some messages every time you plug in or remove a USB device, but you'd have to be switching like crazy to produce 54 GB of messages that way. I suppose a loose cable might make it seem like all the devices are constantly plugged in and removed, but I still don't quite see how the log could grow that big. The excerpt you posted was 3411 characters. Printing all of that once a second for a week would still produce only two gigabytes.
[...]
Finally, an hour or so ago, I tried turning the printer off with its power switch. Since that time, the messages have become fewer, but not stopped.
I'd imagine that the messages about the printer have stopped, and the ones about the keyboard, the mouse and the hub continue. (There's a USB hub inside the KVM switch.) Every time you switch to another machine to look for new messages, you cause more messages.
Right, about both; but some machines get more non-printer messages; I don't know why.
Those messages aren't errors and you don't need to worry about them as long as the log doesn't grow out of control again. It's quite possible that most of those 54 GB was something completely different that hasn't resurfaced yet. I'd recommend doing "ls -l /var/log/messages*" now and then to keep an eye on it, and investigate further if it grows to many megabytes.
I fiddled a bit with the command you suggested (for which, again, many thanks!), and eventually tried doing "$ ls -lh /var/log|less" -- which has the advantage that I need not use a terminal tab logged to root (nor sudo). Doing that on the #1 machine (where I am now) showed a bunch of stuff up to maybe 200K (for very few), and this :
[....] -rw------- 1 root root 26K 2008-10-28 12:37 messages -rw------- 1 root root 299K 2008-10-05 01:00 messages-20081005 -rw------- 1 root root 266K 2008-10-12 03:11 messages-20081012 -rw------- 1 root root 303K 2008-10-19 04:06 messages-20081019 -rw------- 1 root root 0 2008-10-27 11:08 messages-20081026 [....]
I'd like to pipe that into top, or some such, to make it display only the files of 100K and up; but trying to read the man page for top, as usual for powerful commands, makes me think of standing at the foot of a huge cliff of ice.
Somewhere in this thread is a way (or maybe a couple of ways) to skim those files without actually running through them. Maybe I can find it again.
(I've also started skimming through root's mail; and I have to admit a lot more of it makes sense than last time I tried.)
Beartooth wrote:
I'd like to pipe that into top, or some such, to make it display only the files of 100K and up; but trying to read the man page for top, as usual for powerful commands, makes me think of standing at the foot of a huge cliff of ice
How about looking at the largest 30 files and directories there sorted by size in megabytes?
cd /var/log du -ms * | sort -rn | head -n 30
Chris
On Tue, 28 Oct 2008 13:10:15 -0400, Christopher K. Johnson wrote:
Beartooth wrote:
I'd like to pipe that into top, or some such, to make it display only the files of 100K and up; [...]
How about looking at the largest 30 files and directories there sorted by size in megabytes?
cd /var/log du -ms * | sort -rn | head -n 30
Aha! Yes, absolutely. Thank you much. I get :
]# du -ms * | sort -rn | head -n 30 19 audit 1 yum.log 1 Xorg.setup.log 1 Xorg.5.log.old 1 Xorg.5.log 1 Xorg.4.log.old 1 Xorg.4.log 1 Xorg.3.log.old 1 Xorg.3.log 1 Xorg.2.log.old 1 Xorg.2.log 1 Xorg.1.log.old 1 Xorg.1.log 1 Xorg.0.log.old 1 Xorg.0.log 1 wtmp-20081001 1 wtmp 1 setroubleshoot 1 secure-20081026 1 secure-20081019 1 secure-20081012 1 secure-20081005 1 secure 1 samba 1 rpmpkgs-20081026 1 rpmpkgs-20081019 1 rpmpkgs-20081012 1 rpmpkgs-20081005 1 rpmpkgs 1 qtparted-20081025-10h53m26s.log
IOW, nothing like what I was getting yesterday ...
On Tue, 2008-10-28 at 16:53 +0000, Beartooth wrote:
I'd like to pipe that into top, or some such, to make itdisplay only the files of 100K and up; but trying to read the man page for top, as usual for powerful commands, makes me think of standing at the foot of a huge cliff of ice.
'top' for displaying the busiest processes. It has nothing to do with showing file sizes. Maybe you're thinking of
ls -s <files-of-interest>|sort -rn|head -20
or some such.
poc
On Mon, 2008-10-27 at 19:58 +0000, Beartooth wrote:
The printer (an HP PSC 1315v, also USB)) has been in one of those ports for an age. My external DVD RW drive gets plugged into and out of the other port all the time; but the printer stays on the same connection. (The keyboard, mouse (which is indeed an hp), and monitor all get shifted to one individual machine when I do an install or upgrade (so that the monitor and PC can negotiate settings), and otherwise stay on the switch.
I have the exact same printer connected via usb as well, but to my main machine, not a kvm switch. I think switching the switch from one machien to another is causing the messages.
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
On Tue, 28 Oct 2008 02:17:42 -0500, Mike Chambers wrote:
I have the exact same printer connected via usb as well, but to my main machine, not a kvm switch. I think switching the switch from one machien to another is causing the messages.
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
<sigh> "Set it up to share"?? Has that gotten any easier in recent years -- since the last time I gave up trying??
On Tue, 2008-10-28 at 15:16 +0000, Beartooth wrote:
On Tue, 28 Oct 2008 02:17:42 -0500, Mike Chambers wrote:
I have the exact same printer connected via usb as well, but to my main machine, not a kvm switch. I think switching the switch from one machien to another is causing the messages.
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
<sigh> "Set it up to share"?? Has that gotten any easier in recent years -- since the last time I gave up trying??
---- one of my favorite attitudes...defeated before trying
for consideration, it's simple to set up. You connect the printer and configure it to print locally. Then you check the box that tells it to share it with your network.
Then any other linux system on your subnet will automatically recognize and can print to the printer without any configuration at all.
Craig
Beartooth wrote:
On Tue, 28 Oct 2008 02:17:42 -0500, Mike Chambers wrote:
I have the exact same printer connected via usb as well, but to my main machine, not a kvm switch. I think switching the switch from one machien to another is causing the messages.
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
<sigh> "Set it up to share"?? Has that gotten any easier in recent years -- since the last time I gave up trying??
Yes, it has. With a Linux machine host, you just set it up in CUPS. On the rest of the Linux machines, you can have CUPS search the network and it will find the printer. For XP, you will want to install an IPP interface printer. If you really want to, and you have Samba running on the host machine, you can share that way too, but it is more work.
Mikkel
On Tue, 28 Oct 2008 10:58:28 -0500, Mikkel L. Ellertson wrote: [...]
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
<sigh> "Set it up to share"?? Has that gotten any easier in recent years -- since the last time I gave up trying??
Yes, it has. With a Linux machine host, you just set it up in CUPS. On the rest of the Linux machines, you can have CUPS search the network and it will find the printer. For XP, you will want to install an IPP interface printer. If you really want to, and you have Samba running on the host machine, you can share that way too, but it is more work.
Man, it must have! The last time (about FC3 iirc), a friendly soul spent days trying to walk me through it, and finally had to quit before we ever got it to work. After that, I also tried buying a hardware printserver -- and never managed to get that to work, either...
Poking around, I can't tell# CUPS whether what I got into was CUPS or not. $ cups, $ CUPS, # cups, and # CUPS all get "command not found." (rpm -q does tell me I have cups-1.3.9-1.fc9.i386)
So I started looking in the Main Menu. The likeliest launcher, afaict, points to /usr/bin/system-config-printer; is that it? I hit the usual glass wall with that -- asking me things in jargon, as if being English words made their technical sense plain.
I remember there was a way to make at least some browsers handle the configuration -- but not how to launch it; maybe that has gotten easier, too.
Do you have a favorite tutorial on the web somewhere? I know there are some -- which looked a bit daunting last time I saw them...
On Tue, 2008-10-28 at 16:36 +0000, Beartooth wrote:
Poking around, I can't tell# CUPS whether what I got into was CUPS or not. $ cups, $ CUPS, # cups, and # CUPS all get "command not found." (rpm -q does tell me I have cups-1.3.9-1.fc9.i386)
So I started looking in the Main Menu. The likeliest launcher, afaict, points to /usr/bin/system-config-printer; is that it? I hit the usual glass wall with that -- asking me things in jargon, as if being English words made their technical sense plain.
Menu/System/Administration/Printing (which is same as system-config-printer)
You can go both/two ways from the menu I think..
1 - Select on Server/Settings and look for "Publish shared printers..." and check it.
2 - Find your printer, right click on it, and should be a menu with "shared" and a box to check/uncheck.
You should have to save/reactivate or something afterwards. That should get your going or at least close.
Beartooth wrote:
I remember there was a way to make at least some browsers handle the configuration -- but not how to launch it; maybe that has gotten easier, too.
http://localhost:631 I have a bookmark set to it. It is also at the end of the cupsd man page: http://localhost:631/help
Mikkel
Beartooth wrote:
On Tue, 28 Oct 2008 10:58:28 -0500, Mikkel L. Ellertson wrote: [...]
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
<sigh> "Set it up to share"?? Has that gotten any easier in recent years -- since the last time I gave up trying??
Yes, it has. With a Linux machine host, you just set it up in CUPS. On the rest of the Linux machines, you can have CUPS search the network and it will find the printer. For XP, you will want to install an IPP interface printer. If you really want to, and you have Samba running on the host machine, you can share that way too, but it is more work.
Man, it must have! The last time (about FC3 iirc), a friendly soul spent days trying to walk me through it, and finally had to quit before we ever got it to work. After that, I also tried buying a hardware printserver -- and never managed to get that to work, either...
Poking around, I can't tell# CUPS whether what I got into was CUPS or not. $ cups, $ CUPS, # cups, and # CUPS all get "command not found." (rpm -q does tell me I have cups-1.3.9-1.fc9.i386)
So I started looking in the Main Menu. The likeliest launcher, afaict, points to /usr/bin/system-config-printer; is that it? I hit the usual glass wall with that -- asking me things in jargon, as if being English words made their technical sense plain.
Yes, that's it.
I remember there was a way to make at least some browsers handle the configuration -- but not how to launch it; maybe that has gotten easier, too.
You can still do that by pointing a browser at http://localhost:631. That's the administrative interface to CUPS directly, but you really don't need to use it unless you're doing something VERY odd.
system-config-printer (or in Gnome "System->Administration->Printing") it plenty enough.
Do you have a favorite tutorial on the web somewhere? I know there are some -- which looked a bit daunting last time I saw them...
To set up a local printer, click on the "Add Printer" button and enter the data that's asked for. Once it's added, select it in the left pane and on the "Settings" tab, click the "Make Default Printer" button and this new printer becomes the default.
To share local printers, select "Server Settings" in the left pane and check the "Share published printers connected to this system" box. If you want, you can also check "Allow printing from the Internet" box, too.
On the client machines, select "Server Settings" and check the "Show printers shared by other systems" box. After a few minutes, you should see the printers offered by the other machines show up under "Remote Printers" in the left pane. When you ask some application to print, you should be able to choose one of the printers that appear in that left pane.
Can't get a whole lot easier than that. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer ricks@nerd.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - A friend said he climbed to the top of Mount Ranier. My view is - - that if there's no elevator, it must not be that interesting. - ----------------------------------------------------------------------
On Tue, 28 Oct 2008 10:25:00 -0700, Rick Stevens wrote:
I Beartooth wrote:
On Tue, 28 Oct 2008 10:58:28 -0500, Mikkel L. Ellertson wrote: [...]
I have a suggestion....Attach the printer to just one machine, and then set it up to share and then connect each machine to it via the network. That might help a whole lot and cut down on the messages. I hardly ever see those things if at all.
[....]
I remember there was a way to make at least some browsers handle the configuration -- but not how to launch it; maybe that has gotten easier, too.
You can still do that by pointing a browser at http://localhost:631. That's the administrative interface to CUPS directly, but you really don't need to use it unless you're doing something VERY odd.
system-config-printer (or in Gnome "System->Administration->Printing") it plenty enough.
Well, having now looked at present versions of both, I find that the web version seems easier, just in that it gives me a little more idea what it's asking.
To set up a local printer, click on the "Add Printer" button and enter the data that's asked for. Once it's added, select it in the left pane and on the "Settings" tab, click the "Make Default Printer" button and this new printer becomes the default.
To share local printers, select "Server Settings" in the left pane and check the "Share published printers connected to this system" box. If you want, you can also check "Allow printing from the Internet" box, too.
On the client machines, select "Server Settings" and check the "Show printers shared by other systems" box. After a few minutes, you should see the printers offered by the other machines show up under "Remote Printers" in the left pane. When you ask some application to print, you should be able to choose one of the printers that appear in that left pane.
Can't get a whole lot easier than that.
Well, for instance, consider "location." I pulled the USB cable out of the KVM switch and stuck it into my #1 machine. So, if I'm reading between the lines correctly, the same old printer is now "local" to #1, and "remote" to #2 - 4. OK?
Then I go to one of the machines other than #1, and either of those apps (the system-config one OR the web interface) wants me to tell it where the remote printer is, naturally enough. But it gives me no hint, nor any example -- is its location the local IP number of machine #1, or a URI (whatever that is), or what?
So I stumble around a while, by trial and error. Here again, the web interface, being more graphic, gains an advantage -- it's quicker, easier, and surer for me to recognize "Yes, that's the display that seemed to work on the last machine" than it is to do the same the other way.
Beartooth wrote:
Well, for instance, consider "location." I pulled the USB cable out of the KVM switch and stuck it into my #1 machine. So, if I'm reading between the lines correctly, the same old printer is now "local" to #1, and "remote" to #2 - 4. OK?
Yes, but if I remember correctly, the location is the "human" where the printer is located, or anything else. For the location of one printer, I have "Printer in Mike's workshop."
Then I go to one of the machines other than #1, and either of those apps (the system-config one OR the web interface) wants me to tell it where the remote printer is, naturally enough. But it gives me no hint, nor any example -- is its location the local IP number of machine #1, or a URI (whatever that is), or what?
You should not need this with either method - at lease when I go to add a printer in system-config-printers, it first searches for new printers, and should find the new network printer. (I prefer the WEB interface, but both should be working.)
So I stumble around a while, by trial and error. Here again, the web interface, being more graphic, gains an advantage -- it's quicker, easier, and surer for me to recognize "Yes, that's the display that seemed to work on the last machine" than it is to do the same the other way.
Another advantage, if you have CUPS set to allow administration from a machine on the network, is to open a web browser from that machine, and have CUPS on the other machines open - each in their own tab. I also find the help in the web based version easier to use. Then again, one thing about Linux is choice - you can chose the method that works best FOR YOU.
Mikkel
On Tue, 28 Oct 2008 16:17:52 -0500, Mikkel L. Ellertson wrote:
Beartooth wrote:
[....]
You should not need this with either method - at least when I go to add a printer in system-config-printers, it first searches for new printers, and should find the new network printer. (I prefer the WEB interface, but both should be working.)
In case it matters, there is no new printer; only the one they've all been using for months, but through the extra USB port on the KVM switch. [...]
Another advantage, if you have CUPS set to allow administration from a machine on the network, is to open a web browser from that machine, and have CUPS on the other machines open - each in their own tab. I also find the help in the web based version easier to use. Then again, one thing about Linux is choice - you can chose the method that works best FOR YOU.
Wait a minute! After long enough looking at that like a poleaxed ox, I suddenly see a sense it might make. Are you saying I can have four tabs in the same browser *on the same machine* open to CUPS on all four??
I tried substituting some current LAN IP numbers for 127.0.0.1, and got this every time from Firefox :
===== ===== ===== ===== Failed to Connect
Firefox can't establish a connection to the server at 192.168.0.13:631.
Though the site seems valid, the browser was unable to establish a connection.
* Could the site be temporarily unavailable? Try again later. * Are you unable to browse other sites? Check the computer's network connection. * Is your computer or network protected by a firewall or proxy? Incorrect settings can interfere with Web browsing.
===== ===== ===== =====
Does that just mean I have to turn off denyhosts, or something? (I often have troubles with ssh and scp, even with denyhosts turned off on both.) Or are you saying something else altogether?
Beartooth wrote:
On Tue, 28 Oct 2008 16:17:52 -0500, Mikkel L. Ellertson wrote:
Another advantage, if you have CUPS set to allow administration from a machine on the network, is to open a web browser from that machine, and have CUPS on the other machines open - each in their own tab. I also find the help in the web based version easier to use. Then again, one thing about Linux is choice - you can chose the method that works best FOR YOU.
Wait a minute! After long enough looking at that like a poleaxed ox, I suddenly see a sense it might make. Are you saying I can have four tabs in the same browser *on the same machine* open to CUPS on all four??
Yes.
I tried substituting some current LAN IP numbers for 127.0.0.1, and got this every time from Firefox :
You have to turn on remote administration first.
Go to http://localhost:631/admin on each machine - you should only have to do this once. Make sure all these check boxes are checked on the machine with the printer attached:
Share published printers connected to this system Allow printing from the Internet
On every machine that you want to administer from another machine, make sure the "Allow remote administration" box is checked. You may also have to open port 631 in the firewall.
On all the machines that do NOT have the printer attached, make sure the "Show printers shared by other systems" box is checked.
Mikkel
On Wed, 29 Oct 2008 15:11:39 -0500, Mikkel L. Ellertson wrote:
Beartooth wrote:
[...]
Wait a minute! After long enough looking at that like a poleaxed ox, I suddenly see a sense it might make. Are you saying I can have four tabs in the same browser *on the same machine* open to CUPS on all four??
Yes.
I tried substituting some current LAN IP numbers for 127.0.0.1, and got this every time from Firefox :
You have to turn on remote administration first.
Oho the First; I was trying to be too paranoid.
Go to http://localhost:631/admin on each machine - you should only have to do this once. Make sure all these check boxes are checked on the machine with the printer attached:
Share published printers connected to this system Allow printing from the Internet
Oho the Second : that need not mean some script kiddie can run my printer dry printing gibberish??
On every machine that you want to administer from another machine, make sure the "Allow remote administration" box is checked. You may also have to open port 631 in the firewall.
The boxes are checked; lst's hope I don't have to tinker with the firewalls.
On all the machines that do NOT have the printer attached, make sure the "Show printers shared by other systems" box is checked.
Are you saying it should *not* be checked on the one *with* the printer??
Beartooth wrote:
On Wed, 29 Oct 2008 15:11:39 -0500, Mikkel L. Ellertson wrote:
Beartooth wrote:
[...]
Wait a minute! After long enough looking at that like a poleaxed ox, I suddenly see a sense it might make. Are you saying I can have four tabs in the same browser *on the same machine* open to CUPS on all four??
Yes.
I tried substituting some current LAN IP numbers for 127.0.0.1, and got this every time from Firefox :
You have to turn on remote administration first.
Oho the First; I was trying to be too paranoid.
Go to http://localhost:631/admin on each machine - you should only have to do this once. Make sure all these check boxes are checked on the machine with the printer attached:
Share published printers connected to this system Allow printing from the Internet
Oho the Second : that need not mean some script kiddie can run my printer dry printing gibberish??
Not if your firewall is configured correctly, and by that I mean the one facing the internet. If you're not using a router/firewall combo to the internet, then yes, you should add iptables entries that allow your local network access to port 631, but not from the outside.
On every machine that you want to administer from another machine, make sure the "Allow remote administration" box is checked. You may also have to open port 631 in the firewall.
The boxes are checked; lst's hope I don't have to tinker with the firewalls.
On all the machines that do NOT have the printer attached, make sure the "Show printers shared by other systems" box is checked.
Are you saying it should *not* be checked on the one *with* the printer??
Yes. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer ricks@nerd.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - Denial. It ain't just a river in Egypt anymore! - ----------------------------------------------------------------------
On Wed, 29 Oct 2008 14:44:14 -0700, Rick Stevens wrote:
Are you saying it should *not* be checked on the one *with* the printer??
Yes.
OK, I just did. It tells me to hang on while the server restarts. And sits there. When I refresh the tab, it says no changes were made. But if I click to Home and then back to Administration, the box is now unchecked.
So I made the rounds of the other machines, asking for test pages. None made the printer do anything. But the pending lists were breaking two dozen; so I went around once more and cancelled all jobs. (I hope that last was right!)
Beartooth wrote:
On Wed, 29 Oct 2008 15:11:39 -0500, Mikkel L. Ellertson wrote:
Beartooth wrote:
[...]
Wait a minute! After long enough looking at that like a poleaxed ox, I suddenly see a sense it might make. Are you saying I can have four tabs in the same browser *on the same machine* open to CUPS on all four??
Yes.
I tried substituting some current LAN IP numbers for 127.0.0.1, and got this every time from Firefox :
You have to turn on remote administration first.
Oho the First; I was trying to be too paranoid.
Go to http://localhost:631/admin on each machine - you should only have to do this once. Make sure all these check boxes are checked on the machine with the printer attached:
Share published printers connected to this system Allow printing from the Internet
Oho the Second : that need not mean some script kiddie can run my printer dry printing gibberish??
Well, if your local network does not have a firewall between it and the Internet, this is possible. In that case, you probably do not want to share the printer over the network. The text of the checkbox could be a bit clearer - something like over the network instead of over the Internet.
On every machine that you want to administer from another machine, make sure the "Allow remote administration" box is checked. You may also have to open port 631 in the firewall.
The boxes are checked; lst's hope I don't have to tinker with the firewalls.
Well, it is just a checkbox on the firewall configuration GUI, as this is a common port you need open on a local network.
On all the machines that do NOT have the printer attached, make sure the "Show printers shared by other systems" box is checked.
Are you saying it should *not* be checked on the one *with* the printer??
It can be, but unless you have other network printers, or printers attached to other computers that you are sharing, you are better off not having it checked. I tend to be paranoid, and there may be a way to get CUPS into an add printer loop. (Machine 1 shares the printer, machine 2 shares machine 1's printer, and advertises it as a printer available for use on machine 2. machine shares both machine 1's and machines 2's advertised printers, and also advertises them as its available printers, etc...) You should not be able to do it, but someone may manage to do it anyway.
Mikkel
On Wed, 2008-10-29 at 16:52 -0500, Mikkel L. Ellertson wrote:
Well, it is just a checkbox on the firewall configuration GUI, as this is a common port you need open on a local network.
On all the machines that do NOT have the printer attached, make
sure the
"Show printers shared by other systems" box is checked.
Are you saying it should *not* be checked on the one *with*the
printer??
It can be, but unless you have other network printers, or printers attached to other computers that you are sharing, you are better off not having it checked. I tend to be paranoid, and there may be a way to get CUPS into an add printer loop. (Machine 1 shares the printer, machine 2 shares machine 1's printer, and advertises it as a printer available for use on machine 2. machine shares both machine 1's and machines 2's advertised printers, and also advertises them as its available loca ers, etc...) You should not be able to do it, but someone may manage to do it anyway.
In my experience if you don't do something special on a machine with a local printer as well as sharing remote printers. That something special must be done in cupsd file on the machine. Otherwise it will ignore the local printer and only print to the remote printer. -- ======================================================================= Who made the world I cannot tell; 'Tis made, and here am I in hell. My hand, though now my knuckles bleed, I never soiled with such a deed. -- A.E. Housman ======================================================================= Aaron Konstam telephone: (210) 656-0355 e-mail: akonstam@sbcglobal.net
Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems. I see this a lot, because my hardware isn't really up to date enough to handle my monitor optimally. So I tried a couple of tricks that usually help, and they didn't.
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
log files growing at am alarming rate because they are trying to tell you something.
My entire /var partition is only 5GB in size and /var/log is only 660MB of it. That machine has an uptime of 130 days.
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
I can still do what I'm doing now: access the list via Pan on another machine, and go back & forth with my KVM switch; but it makes it hard, for instance, to post things like a list of contents for /var/log, or any whole file in it.
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
Find the particular log file that oversized. Look at it and try to determine *WHY* it is very large and correct the problem that is causing it to grow.
Do you have logrotate installed? Is it rotating log files periodically in order to keep their size down?
On Sun, 26 Oct 2008 17:00:13 -0400, Kevin J. Cummings wrote: [...]
log files growing at am alarming rate because they are trying to tell you something.
That much was one of my guesses; but I have no idea how to find out what. [...]
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
Find the particular log file that oversized. Look at it and try to determine *WHY* it is very large and correct the problem that is causing it to grow.
The first thing I tried was Craig White's suggestion of moving xorg.conf out of the way and rebooting -- because that might have been the quickest way back to pursuing this thread on the problem machine. But it failed; reboot only got me the dread "out of range" message from the monitor.
I'll go try his find command next -- which I'd never have guessed in a thousand years. I'm very glad to have it.
Do you have logrotate installed? Is it rotating log files periodically in order to keep their size down?
According to rpm -q, yes, 3.7.6-5; do I need to take some sort of closer look at it? I did know there was such a thing, and another of my guesses was that it might be misbehaving; but I have no idea how to check that.
On Sun, 2008-10-26 at 20:12 +0000, Beartooth wrote:
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails.
Just by the by, that sort of thing is typical if /tmp isn't writable to, or has the wrong permissions. You *could* have two problemsm /tmp plus your huge log issue.
[tim@gonzales ~]$ ls -ld /tmp/ drwxrwxrwt 32 root root 4096 2008-10-27 15:57 /tmp/
[tim@gonzales ~]$ ls -ldZ /tmp/ drwxrwxrwt root root system_u:object_r:tmp_t:s0 /tmp/
On Mon, 27 Oct 2008 16:01:25 +1030, Tim wrote:
On Sun, 2008-10-26 at 20:12 +0000, Beartooth wrote:
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails.
Just by the by, that sort of thing is typical if /tmp isn't writable to, or has the wrong permissions. You *could* have two problemsm /tmp plus your huge log issue.
[tim@gonzales ~]$ ls -ld /tmp/ drwxrwxrwt 32 root root 4096 2008-10-27 15:57 /tmp/
[root@Hbsk2 btth]# ls -ld /tmp/ drwxrwxrwt 46 root root 4096 2008-10-27 12:07 /tmp/ [root@Hbsk2 btth]#
[tim@gonzales ~]$ ls -ldZ /tmp/ drwxrwxrwt root root system_u:object_r:tmp_t:s0 /tmp/
[root@Hbsk2 btth]# ls -ldZ /tmp/ drwxrwxrwt root root system_u:object_r:tmp_t:s0 /tmp/ [root@Hbsk2 btth]#
[tim@localhost ~]$ uname -r 2.6.26.6-79.fc9.i686
[root@Hbsk2 btth]# uname -r 2.6.26.5-45.fc9.i686 [root@Hbsk2 btth]#
On Sun, Oct 26, 2008 at 12:12 PM, Beartooth Beartooth@swva.net wrote:
At present, the machine offers only a CLI login; if I use is asroot, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
Just a quick note: if either /var or /home are full, X will not start. I think /tmp may also be a problem if full.
~af
On Sun, 2008-10-26 at 21:35 -0800, Aldo Foot wrote:
On Sun, Oct 26, 2008 at 12:12 PM, Beartooth Beartooth@swva.net wrote:
At present, the machine offers only a CLI login; if I use is asroot, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
Just a quick note: if either /var or /home are full, X will not start. I think /tmp may also be a problem if full.
By default, about 5% of the disk space is reserved for root. If free space is low, then a normal user will be out of disk space, but root might still have some of the reserved space. That sounds like it may be the case here.
Dave
On Sun, 26 Oct 2008 21:35:16 -0800, Aldo Foot wrote: [...]
Just a quick note: if either /var or /home are full, X will not start. I think /tmp may also be a problem if full.
df -h says that /dev/mapper/VolGroup00-LogVol00 is at 100%. Normally I can count all the gigs on any of my machines with my fingers, and have fingers left over; nor do I get anywhere near 100% of anything full.
Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems. I see this a lot, because my hardware isn't really up to date enough to handle my monitor optimally. So I tried a couple of tricks that usually help, and they didn't.
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
Um, just a shot in the dark, but it's probably spewing some message over and over. Maybe you should take a look at that.
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
I can still do what I'm doing now: access the list via Pan on another machine, and go back & forth with my KVM switch; but it makes it hard, for instance, to post things like a list of contents for /var/log, or any whole file in it.
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
What should I do first??
Read the error message that's repeated millions of times, and investigate that.
-- Chris
On Mon, 27 Oct 2008 09:58:15 -0400, Chris Snook wrote:
Beartooth wrote:
[...]
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
Um, just a shot in the dark, but it's probably spewing some message over and over. Maybe you should take a look at that.
Sounds only too computer-likely; but how do I look? Maybe if I startx for root, and get into that messages file with nautilus and gedit ...
At present, the machine offers only a CLI login; if I use is as root, startx still works; but if I log in as user, startx fails. So I can't run Pan, and can't c&p directly between anything in it (including this list!) and the CLI.
[...]
Read the error message that's repeated millions of times, and investigate that.
Will do; or at least post it if I do find it. Stay tuned.
Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems. I see this a lot, because my hardware isn't really up to date enough to handle my monitor optimally. So I tried a couple of tricks that usually help, and they didn't.
Then I thought to run baobab. Lo and behold, I was running out of space -- because /var/log shows up consuming 54 GB of space! Yes, fifty- four gigs.
What could be causing this, and what do i do about it??
Probably something's gone wrong and is logging the same message over and over again.
Quick way to find it:
cd /var/log
ls -ltrh
This list all files with the newest file last (your log file was probably the last thing written), and with human readable file sizes. You'll probably see one enormous file. If you don't see it then try the same thing down a directory level
ls -ltrh */*
Once you've found the offending file do tail -500 [whatever file name] to see the last messages written to it, it's probably going to be the same thing over and over again.
Once you know what the message is you can look to fixing it. You can also delete the offending file so your machine can start working again.
Hope this helps
Simon.
On Mon, 27 Oct 2008 14:49:32 +0000, Simon Andrews wrote:
Beartooth wrote:
[....]
What could be causing this, and what do i do about it??
Probably something's gone wrong and is logging the same message over and over again.
Quick way to find it:
cd /var/log
ls -ltrh
This list all files with the newest file last (your log file was probably the last thing written), and with human readable file sizes. You'll probably see one enormous file. If you don't see it then try the same thing down a directory level
ls -ltrh */*
Once you've found the offending file do tail -500 [whatever file name] to see the last messages written to it, it's probably going to be the same thing over and over again.
Once you know what the message is you can look to fixing it. You can also delete the offending file so your machine can start working again.
Hope this helps
As you'll've likely seen by now, I got a bit turned around when suddenly startx no longer worked even for root, and whomped the bloated file with Craig White's big hammer, before somebody gave me a better way than I knew to look at it from the CLI.
So I'm back on the #1 machine, using the whole GUI as user, able to c&p from gnome-terminal to Pan; but the bloat is gone. I *think* that means the info those commands'd've gotten is gone, too.
At any rate, ls -ltrh found nothing bigger than about 300K.
ls -ltrh */* found somewhat bigger ones, but nothing remotely near so horrendous as to run into gigabytes.
[root@Hbsk2 log]# ls -ltrh */* -rw-r--r-- 1 privoxy privoxy 0 2008-09-02 08:34 privoxy/logfile -r-------- 1 root root 5.1M 2008-09-17 16:23 audit/audit.log.3 -r-------- 1 root root 5.1M 2008-09-18 00:21 audit/audit.log.2 -r-------- 1 root root 5.1M 2008-09-18 08:19 audit/audit.log.1 -rw------- 1 root lp 345 2008-09-19 17:31 cups/page_log-20080920 [....] -rw------- 1 root root 728 2008-10-27 11:51 mail/statistics -rw------- 1 root root 3.3M 2008-10-27 12:01 audit/audit.log
samba/old: total 0 [root@Hbsk2 log]#
On Sun, 26 Oct 2008 20:12:21 +0000, Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems.
[....]
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
Having eventually -- thanks to the invaluable help here -- discovered that the culprit was /var/log/messages-20081026, with the whole 54GB; having failed to understand the problem, and finally deleted it just to get back, I'm now on the #1 machine again, and I see :
===== ===== ===== ===== [root@Hbsk2 btth]# df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 72G 18G 51G 26% / /dev/sda1 190M 25M 156M 14% /boot tmpfs 1.5G 48K 1.5G 1% /dev/shm /dev/sdb1 75G 15G 61G 20% /media/System [root@Hbsk2 btth]# ===== ===== ===== =====
All I could tell, from my sadly ineffectual attempts to skim the bad file without GUI means, was that it kept saying something about some USB device, and also that some device couldn't do fax. (I never use fax.)
I'm very sorry not to have a better explanation of what happened, especially inasmuch as I'm also worried. Unless it was some inadvertent command, blamable on my arthritic eyeballs and trifocal fingers, and thus a fluke, it might happen again.
I think logrotate must be working, because both the file for the 25th and the 27th were reported nonexistent. In fact, I'm not at all sure it was me that deleted the bad one; maybe logrotate did it.
I took a look :
===== ===== ===== ===== [root@Hbsk2 btth]# file /var/log/messages-20081027 /var/log/messages-20081027: ERROR: cannot open `/var/log/ messages-20081027' (No such file or directory) [root@Hbsk2 btth]# file /var/log/messages-20081026 /var/log/messages-20081026: empty [root@Hbsk2 btth]# file /var/log/messages-20081025 /var/log/messages-20081025: ERROR: cannot open `/var/log/ messages-20081025' (No such file or directory) [root@Hbsk2 btth]# ===== ===== ===== =====
Does that tell anyone anything more than it does me? Does it look like, or at least compatible with, logrotate?
I suppose I should take a look at root's mail. What should I look there for? (I think I do have logwatch running on this machine.)
On Mon, 2008-10-27 at 15:50 +0000, Beartooth wrote:
On Sun, 26 Oct 2008 20:12:21 +0000, Beartooth wrote:
One of my F9 machines -- actually my #1 main machine -- started having display problems.
[....]
Obviously, I want to cut /var/log with electronic double-bitted axes in both hands, to get to my user's GUI again; and to find the source of the bloat and correct it.
Having eventually -- thanks to the invaluable help here -- discovered that the culprit was /var/log/messages-20081026, with the whole 54GB; having failed to understand the problem, and finally deleted it just to get back, I'm now on the #1 machine again, and I see :
===== ===== ===== ===== [root@Hbsk2 btth]# df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 72G 18G 51G 26% / /dev/sda1 190M 25M 156M 14% /boot tmpfs 1.5G 48K 1.5G 1% /dev/shm /dev/sdb1 75G 15G 61G 20% /media/System [root@Hbsk2 btth]# ===== ===== ===== =====
All I could tell, from my sadly ineffectual attempts to skim the bad file without GUI means, was that it kept saying something about some USB device, and also that some device couldn't do fax. (I never use fax.)
I'm very sorry not to have a better explanation of what happened, especially inasmuch as I'm also worried. Unless it was some inadvertent command, blamable on my arthritic eyeballs and trifocal fingers, and thus a fluke, it might happen again.
I think logrotate must be working, because both the file for the 25th and the 27th were reported nonexistent. In fact, I'm not at all sure it was me that deleted the bad one; maybe logrotate did it.
I took a look :
===== ===== ===== ===== [root@Hbsk2 btth]# file /var/log/messages-20081027 /var/log/messages-20081027: ERROR: cannot open `/var/log/ messages-20081027' (No such file or directory) [root@Hbsk2 btth]# file /var/log/messages-20081026 /var/log/messages-20081026: empty [root@Hbsk2 btth]# file /var/log/messages-20081025 /var/log/messages-20081025: ERROR: cannot open `/var/log/ messages-20081025' (No such file or directory) [root@Hbsk2 btth]# ===== ===== ===== =====
Does that tell anyone anything more than it does me? Does it look like, or at least compatible with, logrotate?
I suppose I should take a look at root's mail. What should I look there for? (I think I do have logwatch running on this machine.)
---- probably because you're out of space to create them
Just do this at command line (as root)
# > /var/log/messages
that will completely empty it out (the # sign signifies root and prevents e-mail from thinking it is part of a reply. Don't type the # sign)
Craig
-
Aaron Konstam -
Aldo Foot -
Beartooth -
Björn Persson -
Chris Snook -
Christopher K. Johnson -
Craig White -
David G. Mackay -
g -
Jonathan Dieter -
Kevin Cummings -
Mike Chambers -
Mikkel -
Patrick O'Callaghan -
Rick Stevens -
Simon Andrews -
Tim