My firefox on F38 freezes a lot. I get "Firefox" is not responding a lot. At one time, I thought it was because of too many windows and tabs open. I closed several windows. Shutting it down and bringing it back up often helped, but not necessarily for very long. According to top, firefox freezes even when the load average is less than two, most of swap is free and neither firefox nor its minions are near the top of top. My new suspicion is that it waits for something that does not happen often enough. How do I diagnose this?
On 11/01/2023 09:12 PM, Michael Hennebry wrote:
My firefox on F38 freezes a lot. I get "Firefox" is not responding a lot.
Firefox can be quite a memory hog, depending on what you're doing. What does
free -h
show you, as well as
df
You might be running short of memory and/or disk space.
On Wed, 2023-11-01 at 21:24 -0600, Joe Zeff wrote:
Firefox can be quite a memory hog, depending on what you're doing. What does
free -h
show you, as well as
df
There's an about:memory page in Firefox, an about:performance, an about:processes and a bunch of other info listed if you type about:about into its address bar.
I had firefox continually freezing (Xorg) taking the gui with it. I then aliased firefox to: cpulimit -i -l 200 /usr/bin/firefox "$@" and it stopped taking the rest of the gui with it. Recently I (had to) switched to Wayland and firefox (with the alias) has only crashed 3 or 4 times but it has not taken the rest of the gui with it.
On 11/2/23 01:36, Tim via users wrote:
On Wed, 2023-11-01 at 21:24 -0600, Joe Zeff wrote:
Firefox can be quite a memory hog, depending on what you're doing. What does
free -h
show you, as well as
df
There's an about:memory page in Firefox, an about:performance, an about:processes and a bunch of other info listed if you type about:about into its address bar.
Thank you Joe Z, Tim and Richard E. Firefox just starting freezing. [hennebry@fedora ~]$ df -h Filesystem Size Used Avail Use% Mounted on devtmpfs 4.0M 0 4.0M 0% /dev tmpfs 3.9G 0 3.9G 0% /dev/shm tmpfs 1.6G 1.7M 1.6G 1% /run /dev/sda3 31G 7.1G 23G 25% / tmpfs 3.9G 288K 3.9G 1% /tmp /dev/sda6 5.9G 4.4G 1.2G 80% /var /dev/sda7 32G 23G 7.1G 76% /home /dev/sdb1 30G 8.1G 20G 30% /run/media/hennebry/data3 tmpfs 785M 11M 774M 2% /run/user/1000 [hennebry@fedora ~]$ free -h total used free shared buff/cache available Mem: 7.7Gi 2.9Gi 2.5Gi 472Mi 2.3Gi 4.0Gi Swap: 7.7Gi 93Mi 7.6Gi
Note that I do not have disk swap. I expect the above is at least partly a computation involving zswap. about:performance takes me to about:processes . Main, google and yahoo mail tabs seem to be taking the most CPU and memory, 20% 618MB, 0.6% 411MG and 0.18% 159MB . The latter two occasionally flicker to about 20% CPU. No smoking gun.
Memory report: Main Process (pid 460754) Explicit Allocations 1,224.93 MB (100.0%) -- explicit ... 682.46 MB (100682.46 MB (100.0%) -- gfx ... 1,148.72 MB (100.0%) -- heap-committed
3.47 MB (100.0%) -- images ... 227 (100.0%) -- ipc-channels ... 635 (100.0%) -- ipc-channels-peak ... 2 (100.0%) -- js-helper-threads ... 101.68 MB (100.0%) -- js-main-runtime ... 52.69 MB (100.0%) -- js-main-runtime-gc-heap-committed ...
37 (100.0%) -- js-main-runtime-realms ... 103 (100.0%) -- message-manager ... 2,581 (100.0%) -- observer-service ... 894 (100.0%) -- observer-service-suspect ... 1,145 (100.0%) -- preference-service ... 0 (100.0%) -- queued-ipc-messages ... 0.14 MB (100.0%) -- shared-string-bundles 41.88 MB (100.0%) ++ window-objects ...
Apparently there is a tree for each tab. I have a lot of tabs.
This took me about an hour. When firefox is frozen, I cannot scroll it nor copy from it. I have to wait for a good second.
Firefox does not take gnome with it.
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 10:21 AM, Michael Hennebry wrote:
/dev/sda7 32G 23G 7.1G 76% /home
If I'm not mistaken, firefox puts its working files both in /tmp and in your home directory. If so, that may be your problem. Consider installing Bleachbit and letting it clean all of the cruft out.
According to du, ~/.mozilla contains 887M .
On 11/02/2023 01:55 PM, Michael Hennebry wrote:
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 10:21 AM, Michael Hennebry wrote:
/dev/sda7 32G 23G 7.1G 76% /home
If I'm not mistaken, firefox puts its working files both in /tmp and in your home directory. If so, that may be your problem. Consider installing Bleachbit and letting it clean all of the cruft out.
According to du, ~/.mozilla contains 887M .
Is that with Firefox running or closed? In either case, my suggestion of Bleachbit still stands.
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 01:55 PM, Michael Hennebry wrote:
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 10:21 AM, Michael Hennebry wrote:
/dev/sda7 32G 23G 7.1G 76% /home
If I'm not mistaken, firefox puts its working files both in /tmp and in your home directory. If so, that may be your problem. Consider installing Bleachbit and letting it clean all of the cruft out.
According to du, ~/.mozilla contains 887M .
Is that with Firefox running or closed? In either case, my suggestion of Bleachbit still stands.
With firefox frozen.
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 02:10 PM, Michael Hennebry wrote:
With firefox frozen.
And with it closed? What I'm trying to find out is how much of that is stuff Firefox needs to keep between sessions and how much of it is session specific.
I'll get back to you on that, after I get the requested memory data. about:performances seems to be treate as an alias for about:processes .
On Thu, 2 Nov 2023, Joe Zeff wrote:
On 11/02/2023 02:10 PM, Michael Hennebry wrote:
With firefox frozen.
And with it closed? What I'm trying to find out is how much of that is stuff Firefox needs to keep between sessions and how much of it is session specific.
885M with firefox closed.
BTW I've not the when the not responding message appears, I cannot even raise the window, but when the message is gone, I can. Usually I cannot even click on the wait option, not that it matters.
On Thu, 2023-11-02 at 11:21 -0500, Michael Hennebry wrote:
Main, google and yahoo mail tabs seem to be taking the most CPU and memory, 20% 618MB, 0.6% 411MG and 0.18% 159MB . The latter two occasionally flicker to about 20% CPU. No smoking gun.
It's well to remember that it can be a cumulative problem. Not always one thing gone bad, but a bunch of things making life difficult.
Sometimes it's the silliest of things that make a system drag. In my case, I've found the live chat on a youtube stream to be a big CPU hog. It doesn't appear to be doing much, but there's probably some inefficiency in how they did it. And if your PC has to start paging to deal with it, that can really bring a system to its knees.
Dynamic pages that update themselves are going to be the worst. Whether that being status changes, or continually loading advertising. They're not always written well, make assumptions that the entire computer is at their disposal, that all users have a super fast CPU and infinite RAM, and often entail a large conglomeration of scripts. You may find a script blocker, or advert blocker, makes a huge difference to using such sites.
On Wed, 1 Nov 2023 22:12:14 -0500 (CDT) Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
My firefox on F38 freezes a lot. I get "Firefox" is not responding a lot. At one time, I thought it was because of too many windows and tabs open. I closed several windows. Shutting it down and bringing it back up often helped, but not necessarily for very long. According to top, firefox freezes even when the load average is less than two, most of swap is free and neither firefox nor its minions are near the top of top. My new suspicion is that it waits for something that does not happen often enough. How do I diagnose this?
You could try the troubleshoot mode, https://support.mozilla.org/en-US/kb/diagnose-firefox-issues-using-troublesh...
yeah...! I also had the same symptoms... it wasn't good.
firefox-118.0.2.tar.bz2 https://support.mozilla.org/en-US/kb/install-firefox-linux
I manually setup fedora38 and applied it through the link process, and it seems to have improved,
so I'm using it well.
type this In the address bar: about:performance click on the memory column and sort by biggest first. see what specific web pages are using excessive ram (usually close to 1gb or more). click on the tab: entryh that sucks and a X will appear on the far right, and click the X and the specific tab gets killed.. weather.com and reddit.com are the ones that typically misbehave for me.
I think firefox needs a simple feature that basically says if any tab uses X kill/restart it.
On Wed, Nov 1, 2023 at 10:12 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
My firefox on F38 freezes a lot. I get "Firefox" is not responding a lot. At one time, I thought it was because of too many windows and tabs open. I closed several windows. Shutting it down and bringing it back up often helped, but not necessarily for very long. According to top, firefox freezes even when the load average is less than two, most of swap is free and neither firefox nor its minions are near the top of top. My new suspicion is that it waits for something that does not happen often enough. How do I diagnose this?
-- Michael hennebry@mail.cs.ndsu.NoDak.edu "I was just thinking about the life of a pumpkin. Grow up in the sun, happily entwined with others, and then someone comes along, cuts you open, and rips your guts out." -- Buffy _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Thu, 2 Nov 2023, Roger Heflin wrote:
type this In the address bar: about:performance
I cannot get to about:performance . 'Tis not in the about:about list. When I type about:performance , I get about:memory .
click on the memory column and sort by biggest first. see what specific web pages are using excessive ram (usually close to 1gb or more). click on the tab: entryh that sucks and a X will appear on the far right, and click the X and the specific tab gets killed.. weather.com and reddit.com are the ones that typically misbehave for me.
I think firefox needs a simple feature that basically says if any tab uses X kill/restart it.
It does seem that about:processes and about:performance seem to be same page.
So yes about:process. When I googled it pointed me to about:performance so I have been using that without noticing that it was an alias.
finding and killing/restarting the specific tab has been working for me for a couple of months.
On Thu, Nov 2, 2023 at 4:49 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Thu, 2 Nov 2023, Roger Heflin wrote:
type this In the address bar: about:performance
I cannot get to about:performance . 'Tis not in the about:about list. When I type about:performance , I get about:memory .
click on the memory column and sort by biggest first. see what specific web pages are using excessive ram (usually close to 1gb or more). click on the tab: entryh that sucks and a X will appear on the far right, and click the X and the specific tab gets killed.. weather.com and reddit.com are the ones that typically misbehave for me.
I think firefox needs a simple feature that basically says if any tab uses X kill/restart it.
-- Michael hennebry@mail.cs.ndsu.NoDak.edu "I was just thinking about the life of a pumpkin. Grow up in the sun, happily entwined with others, and then someone comes along, cuts you open, and rips your guts out." -- Buffy _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
here are the top memory entries from about:processes: 872 firefox 555 weather.com 254 imasdk.googleapis.com 210 linkedin.com 204 Extensions 146 yahoo.com 132 avrfreaks.net 116 googlesyndication.com 88 yahoo.net 85 hackerrank.com 63 linkedin.com 37 github.com
I note weather.com, as was previously suggested as a possible miscreant.
On Fri, 3 Nov 2023, Michael Hennebry wrote:
here are the top memory entries from about:processes: 872 firefox 555 weather.com 254 imasdk.googleapis.com 210 linkedin.com 204 Extensions 146 yahoo.com 132 avrfreaks.net 116 googlesyndication.com 88 yahoo.net 85 hackerrank.com 63 linkedin.com 37 github.com
I note weather.com, as was previously suggested as a possible miscreant.
Replacing weather.com with start.fedoraproject.com did not seem to help.
\
On Fri, Nov 3, 2023 at 12:57 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Fri, 3 Nov 2023, Michael Hennebry wrote:
here are the top memory entries from about:processes: 872 firefox 555 weather.com 254 imasdk.googleapis.com 210 linkedin.com 204 Extensions 146 yahoo.com 132 avrfreaks.net 116 googlesyndication.com 88 yahoo.net 85 hackerrank.com 63 linkedin.com 37 github.com
I note weather.com, as was previously suggested as a possible miscreant.
Replacing weather.com with start.fedoraproject.com did not seem to help.
How much ram does the machine firefox is running on have?
On Fri, 3 Nov 2023, Roger Heflin wrote:
How much ram does the machine firefox is running on have?
8 G
$ free -h total used free shared buff/cache available Mem: 7.7Gi 4.1Gi 882Mi 572Mi 2.7Gi 2.7Gi Swap: 7.7Gi 98Mi 7.6Gi $ I have no swap partition.
On Fri, 3 Nov 2023, Joe Zeff wrote:
On 11/03/2023 11:50 AM, Michael Hennebry wrote:
210 linkedin.com
146 yahoo.com 88 yahoo.net
63 linkedin.com
Why do you have both an extension and a tab for linkedin.com, and extensions for both yahoo.com and yahoo.net?
I have more than one linkedin tab. As for yahoo.com and yahoo.net, I have no idea.
On Wed, Nov 1, 2023 at 11:12 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
My firefox on F38 freezes a lot. I get "Firefox" is not responding a lot. At one time, I thought it was because of too many windows and tabs open. I closed several windows. Shutting it down and bringing it back up often helped, but not necessarily for very long. According to top, firefox freezes even when the load average is less than two, most of swap is free and neither firefox nor its minions are near the top of top. My new suspicion is that it waits for something that does not happen often enough. How do I diagnose this?
I don't know how to diagnose Firefox problems. Others have given you some places to go.
But I've found most of the time, my problems were related to GPU drivers and acceleration. Here's how to disable them: https://support.mozilla.org/en-US/kb/upgrade-graphics-drivers-use-hardware-acceleration#w_turning-off-hardware-acceleration.
Jeff
By now, I am fairly sure that the reason firefox freezes on me is neither a lack of memory nor a lack of CPU. My inference is that it is waiting for something. How do I discover what?
On Thu, Nov 23, 2023 at 2:21 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
By now, I am fairly sure that the reason firefox freezes on me is neither a lack of memory nor a lack of CPU. My inference is that it is waiting for something. How do I discover what?
I would investigate the virtual memory system. That's based on your reply to Roger Heflin, and the 8GB of RAM and no swap file. Maybe start with `vm.overcommit_memory = 2` in `/etc/sysctl.conf`. The 2 says, "say no if we don't have the memory".
Firefox is a memory hog. It gets worse as you add more tabs. I would add more RAM and a swap partition. Or I would use a less resource intensive browser.
If you really want to see memory pressure pain, then install Solaris. It does not overcommit memory like Linux does.
Jeff
On Thu, 23 Nov 2023, Jeffrey Walton wrote:
On Thu, Nov 23, 2023 at 2:21?PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
By now, I am fairly sure that the reason firefox freezes on me is neither a lack of memory nor a lack of CPU. My inference is that it is waiting for something. How do I discover what?
I would investigate the virtual memory system. That's based on your reply to Roger Heflin, and the 8GB of RAM and no swap file. Maybe
You mean this: Michael Hennebry wrote (some whitespace deleted):
$ free -h total used free shared buff/cache available Mem: 7.7Gi 4.1Gi 882Mi 572Mi 2.7Gi 2.7Gi Swap: 7.7Gi 98Mi 7.6Gi $ I have no swap partition.
? I'd have thought 882Mi free and 2.7Gi available would be good. BTW I have a lot of tabs, but not a lot of videos. Mostly my tabs are things to read.
start with `vm.overcommit_memory = 2` in `/etc/sysctl.conf`. The 2 says, "say no if we don't have the memory".
From the persistence of the belief that I am running out of memory, I infer one of two scenarios is assumed: 1: firefox is waiting on memory is was told it has, but might never get. 2: firefox has been told memory is unavailable, but does not do anything sensible with that information. The output from free would seem to preclude both scenarios.
What is the state of a process that is waiting on commited, but unavailable, memory?
On 11/23/23 16:53, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Jeffrey Walton wrote:
On Thu, Nov 23, 2023 at 2:21?PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
By now, I am fairly sure that the reason firefox freezes on me is neither a lack of memory nor a lack of CPU. My inference is that it is waiting for something. How do I discover what?
I would investigate the virtual memory system. That's based on your reply to Roger Heflin, and the 8GB of RAM and no swap file. Maybe
You mean this: Michael Hennebry wrote (some whitespace deleted):
$ free -h total used free shared buff/cache available Mem: 7.7Gi 4.1Gi 882Mi 572Mi 2.7Gi 2.7Gi Swap: 7.7Gi 98Mi 7.6Gi $ I have no swap partition.
? I'd have thought 882Mi free and 2.7Gi available would be good. BTW I have a lot of tabs, but not a lot of videos. Mostly my tabs are things to read.
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
On Thu, 23 Nov 2023, Samuel Sieb wrote:
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
journalctl and then what?
On 11/23/23 17:12, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
journalctl and then what?
I used "journalctl -r" which will display lines reversed starting from the most recent. Or you can run "journalctl -b" and press "G" to go to the end. It uses the same navigation keys as "less".
On Thu, 23 Nov 2023, Samuel Sieb wrote:
On 11/23/23 17:12, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
journalctl and then what?
I used "journalctl -r" which will display lines reversed starting from the most recent. Or you can run "journalctl -b" and press "G" to go to the end. It uses the same navigation keys as "less".
I used journalctl -r -g irefox Rather a lot of ... [Child 287536, MediaDecoderStateMachine #1] WARNING: Decoder=7fde2015e400 Decode error:... I suspect they relate to a video I was trying to watch. Did not see anything that seemed to be a memory thing.
Time for supper.
On 11/23/23 17:37, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
On 11/23/23 17:12, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
journalctl and then what?
I used "journalctl -r" which will display lines reversed starting from the most recent. Or you can run "journalctl -b" and press "G" to go to the end. It uses the same navigation keys as "less".
I used journalctl -r -g irefox Rather a lot of ... [Child 287536, MediaDecoderStateMachine #1] WARNING: Decoder=7fde2015e400 Decode error:... I suspect they relate to a video I was trying to watch. Did not see anything that seemed to be a memory thing.
I'm saying it wasn't a memory thing. Here are some example lines that I saw. They were all together at the time that firefox was stuck.
rtkit-daemon[1090]: Successfully made thread 7825 of process 6769 (/usr/lib64/firefox/firefox) owned by '1000' RT at priority 10.
firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: g_dbus_connection_unregister_object: assertion 'G_IS_DBUS_CONNECTION (connection)' failed: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: g_dbus_connection_unregister_object: assertion 'G_IS_DBUS_CONNECTION (connection)' failed firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: Error releasing name org.mpris.MediaPlayer2.firefox.instance2327: Timeout was reached: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: Error releasing name org.mpris.MediaPlayer2.firefox.instance2327: Timeout was reached firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: Create input context failed: Timeout was reached.: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: Create input context failed: Timeout was reached.
On Thu, 23 Nov 2023, Samuel Sieb wrote:
I'm saying it wasn't a memory thing. Here are some example lines that I saw. They were all together at the time that firefox was stuck.
rtkit-daemon[1090]: Successfully made thread 7825 of process 6769 (/usr/lib64/firefox/firefox) owned by '1000' RT at priority 10.
firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: g_dbus_connection_unregister_object: assertion 'G_IS_DBUS_CONNECTION (connection)' failed: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: g_dbus_connection_unregister_object: assertion 'G_IS_DBUS_CONNECTION (connection)' failed firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: Error releasing name org.mpris.MediaPlayer2.firefox.instance2327: Timeout was reached: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
I got eleven of those yesterday. file /builddir/build/BUILD/firefox-119.0/toolkit/xre/nsSigHandlers.cpp:167
firefox[2327]: Error releasing name org.mpris.MediaPlayer2.firefox.instance2327: Timeout was reached firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: Create input context failed: Timeout was reached.: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: Create input context failed: Timeout was reached.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 11/23/23 17:37, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
On 11/23/23 17:12, Michael Hennebry wrote:
On Thu, 23 Nov 2023, Samuel Sieb wrote:
Have you checked the journal? I had this happen on a computer that definitely wasn't low on memory and the journal had a bunch of messages from firefox about timeouts.
journalctl and then what?
I used "journalctl -r" which will display lines reversed starting from the most recent. Or you can run "journalctl -b" and press "G" to go to the end. It uses the same navigation keys as "less".
I used journalctl -r -g irefox Rather a lot of ... [Child 287536, MediaDecoderStateMachine #1] WARNING: Decoder=7fde2015e400 Decode error:... I suspect they relate to a video I was trying to watch. Did not see anything that seemed to be a memory thing.
Possibly related, I'm also getting issues where typing doesn't work for a long time with the following messages in the log:
firefox.desktop[2327]: [Parent 2327, Main Thread] WARNING: Process Key Event failed: Timeout was reached.: 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:187
firefox[2327]: Process Key Event failed: Timeout was reached.
chromium is freezing on me now, too. journalctl -r -g romium does not reveal any errors or warnings.
On Fri, Nov 24, 2023 at 8:57 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
chromium is freezing on me now, too. journalctl -r -g romium does not reveal any errors or warnings.
-- Michael hennebry@mail.cs.ndsu.NoDak.edu "His longhand was fairly good in his youth, but as he got older it got smaller, more scribbly, and harder to read; although, like being hanged, one can get used to it." -- Gordon Dickson on H.P. Lovecraft's handwriting
You are having far too many problems with too many programs. The common fixes are not helping.
Try new hardware.
Jeff
On Fri, 2023-11-24 at 21:02 -0500, Jeffrey Walton wrote:
You are having far too many problems with too many programs. The common fixes are not helping.
Remembering tales from long ago - how updating an OS often seemed to induce faults in hardware that was apparently working fine before. There's many reasons why this may be the case.
The newer code is poking different areas of memory than before. Previously a memory fault existed, but not in an area that you noticed (maybe one pixel in the display never worked) and the system didn't throw a tantrum about, but now that bit of RAM is being used by some important operating system process. I've experienced this kind of thing (one install worked fine, apparently, even with RAM that tested as faulty, the next install did not). And, of course, there's things like manufacturers releasing faulty hardware that fails on a newer driver thanks to happenstance that the prior driver didn't provoke.
The install process was CPU intensive for a prolonged time and stressed something on a system that has spent most of its time the last few years just idling only. A heatsink wasn't good enough, or clogged up with fluff, or not fastened properly, or the thermal paste dried out or wasn't applied well in the first place, or the cooling fan was failing, and the CPU overheated. Some power filtering component on the edge of dying, did. A vibrating DVD drive finally weakened some bad soldering joint...
And another one I've experienced: Prior to an upgrade, you've picked up the box and cleaned it, or just moved it. The box isn't rigid, it twists a bit, and some of the daughter boards aren't properly seated to the motherboard any more (video cards, memory cards, etc). This also happens when the ambient temperature changes a lot - things creep out. Even a CPU socket can go bad, and I'm surprised this doesn't happen more often, considering the huge cooling devices that are crushed onto them, these days. As far back as the old Apple ][ it was a common debugging procedure to re-seat all the cards and socketed chips, to fix odd failures. Also, some cards didn't have very good edge connectors, cleaning them often made things better.
I remember a range of Macs had problems (as they aged) with the factory soldering not being up to par. People would remove all the plastic parts and back the motherboard in an effort to re-flow the solder. I was given one in that condition, but decided it wasn't worth the pain to try and fix. Solder faults on complex machine boards tend to be everywhere, not just one spot. Often caused by the solder being too contaminated on the build day, or wrong temperature, or the boards being contaminated. Things sometimes just age and fail from cumulative decay.
So, come OS install and update times, I tend to open a box, inspect heatsinks, clean it, and reseat all the connections. Having a spare power supply to swap over is handy, too. They don't always age well, especially the bargain basement types.
On Sat, Nov 25, 2023 at 2:08 AM Tim ignored_mailbox@yahoo.com.au wrote:
On Fri, 2023-11-24 at 21:02 -0500, Jeffrey Walton wrote:
You are having far too many problems with too many programs. The common fixes are not helping.
Remembering tales from long ago - how updating an OS often seemed to induce faults in hardware that was apparently working fine before. There's many reasons why this may be the case.
The newer code is poking different areas of memory than before. Previously a memory fault existed, but not in an area that you noticed (maybe one pixel in the display never worked) and the system didn't throw a tantrum about, but now that bit of RAM is being used by some important operating system process. I've experienced this kind of thing (one install worked fine, apparently, even with RAM that tested as faulty, the next install did not). And, of course, there's things like manufacturers releasing faulty hardware that fails on a newer driver thanks to happenstance that the prior driver didn't provoke.
The install process was CPU intensive for a prolonged time and stressed something on a system that has spent most of its time the last few years just idling only. A heatsink wasn't good enough, or clogged up with fluff, or not fastened properly, or the thermal paste dried out or wasn't applied well in the first place, or the cooling fan was failing, and the CPU overheated. Some power filtering component on the edge of dying, did. A vibrating DVD drive finally weakened some bad soldering joint...
And another one I've experienced: Prior to an upgrade, you've picked up the box and cleaned it, or just moved it. The box isn't rigid, it twists a bit, and some of the daughter boards aren't properly seated to the motherboard any more (video cards, memory cards, etc). This also happens when the ambient temperature changes a lot - things creep out. Even a CPU socket can go bad, and I'm surprised this doesn't happen more often, considering the huge cooling devices that are crushed onto them, these days. As far back as the old Apple ][ it was a common debugging procedure to re-seat all the cards and socketed chips, to fix odd failures. Also, some cards didn't have very good edge connectors, cleaning them often made things better.
I remember a range of Macs had problems (as they aged) with the factory soldering not being up to par. People would remove all the plastic parts and back the motherboard in an effort to re-flow the solder. I was given one in that condition, but decided it wasn't worth the pain to try and fix. Solder faults on complex machine boards tend to be everywhere, not just one spot. Often caused by the solder being too contaminated on the build day, or wrong temperature, or the boards being contaminated. Things sometimes just age and fail from cumulative decay.
So, come OS install and update times, I tend to open a box, inspect heatsinks, clean it, and reseat all the connections. Having a spare power supply to swap over is handy, too. They don't always age well, especially the bargain basement types.
I'm guessing he's got a chip that is overheating when powered on or under load. Maybe due to a bad fan, maybe a bad capacitor, maybe something else.
It would be interesting to see an IR image of the motherboard and chips while powered on. See if anything is hitting 70 or 80°C.
Jeff
Only if freezing includes the machine crashing. If the machine is not crashing and it recovers from the freeze without a reboot/power cycle then we are back to memory/paging being a problem. Note I have seen swapping act as badly. And it may only be every so often that it runs out of ram. My older 10GB machine was having serious issues with ram usage a couple of years ago until I retired it (upgrading the ram would have cost 50% of the price of a new much faster machine--ddr2 4gb dimms are expensive).
Hardware issues do not typically leave the machine up (except for intel/amd throttling down when getting hot, but even that will crash if it gets bad enough). install kernel-tools and run turbostat it will show the cpu freqs and cpu temps.
On Sat, Nov 25, 2023 at 4:01 AM Jeffrey Walton noloader@gmail.com wrote:
On Sat, Nov 25, 2023 at 2:08 AM Tim ignored_mailbox@yahoo.com.au wrote:
On Fri, 2023-11-24 at 21:02 -0500, Jeffrey Walton wrote:
You are having far too many problems with too many programs. The common fixes are not helping.
Remembering tales from long ago - how updating an OS often seemed to induce faults in hardware that was apparently working fine before. There's many reasons why this may be the case.
The newer code is poking different areas of memory than before. Previously a memory fault existed, but not in an area that you noticed (maybe one pixel in the display never worked) and the system didn't throw a tantrum about, but now that bit of RAM is being used by some important operating system process. I've experienced this kind of thing (one install worked fine, apparently, even with RAM that tested as faulty, the next install did not). And, of course, there's things like manufacturers releasing faulty hardware that fails on a newer driver thanks to happenstance that the prior driver didn't provoke.
The install process was CPU intensive for a prolonged time and stressed something on a system that has spent most of its time the last few years just idling only. A heatsink wasn't good enough, or clogged up with fluff, or not fastened properly, or the thermal paste dried out or wasn't applied well in the first place, or the cooling fan was failing, and the CPU overheated. Some power filtering component on the edge of dying, did. A vibrating DVD drive finally weakened some bad soldering joint...
And another one I've experienced: Prior to an upgrade, you've picked up the box and cleaned it, or just moved it. The box isn't rigid, it twists a bit, and some of the daughter boards aren't properly seated to the motherboard any more (video cards, memory cards, etc). This also happens when the ambient temperature changes a lot - things creep out. Even a CPU socket can go bad, and I'm surprised this doesn't happen more often, considering the huge cooling devices that are crushed onto them, these days. As far back as the old Apple ][ it was a common debugging procedure to re-seat all the cards and socketed chips, to fix odd failures. Also, some cards didn't have very good edge connectors, cleaning them often made things better.
I remember a range of Macs had problems (as they aged) with the factory soldering not being up to par. People would remove all the plastic parts and back the motherboard in an effort to re-flow the solder. I was given one in that condition, but decided it wasn't worth the pain to try and fix. Solder faults on complex machine boards tend to be everywhere, not just one spot. Often caused by the solder being too contaminated on the build day, or wrong temperature, or the boards being contaminated. Things sometimes just age and fail from cumulative decay.
So, come OS install and update times, I tend to open a box, inspect heatsinks, clean it, and reseat all the connections. Having a spare power supply to swap over is handy, too. They don't always age well, especially the bargain basement types.
I'm guessing he's got a chip that is overheating when powered on or under load. Maybe due to a bad fan, maybe a bad capacitor, maybe something else.
It would be interesting to see an IR image of the motherboard and chips while powered on. See if anything is hitting 70 or 80°C.
Jeff
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Sat, 25 Nov 2023, Roger Heflin wrote:
Hardware issues do not typically leave the machine up (except for intel/amd throttling down when getting hot, but even that will crash if it gets bad enough). install kernel-tools and run turbostat it will show the cpu freqs and cpu temps.
My CPU is Intel. I'll try kernel-tools.
On Mon, 27 Nov 2023, Michael Hennebry wrote:
On Sat, 25 Nov 2023, Roger Heflin wrote:
Hardware issues do not typically leave the machine up (except for intel/amd throttling down when getting hot, but even that will crash if it gets bad enough). install kernel-tools and run turbostat it will show the cpu freqs and cpu temps.
My CPU is Intel. I'll try kernel-tools.
Today is the first time firefox froze on me since leaving turbostat running.
It is almost certainly going to be a memory issue.
In fact earlier today Firefox made my 32G windows 11 device mostly useless (not a complete lockup for anything but firefox, but everything else was horribly slow).
I have had to kill firefox on multiple different websites and multiple different machines. On ones with a lot of ram they don't lockup, but on lower ram devices (16G and under) they start paging and become otherwise useless and lock up enough that the power off button is the fastest solution.
If you have a second device that you can ssh into the first device, ssh into the first device and run top and leave it running. On lockup see what top last said before you reset/powered it off.
You could also do top > top.out & and simply leave that running and the go see what top was showing next lockup.
On Mon, Dec 11, 2023 at 5:13 PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Mon, 27 Nov 2023, Michael Hennebry wrote:
On Sat, 25 Nov 2023, Roger Heflin wrote:
Hardware issues do not typically leave the machine up (except for intel/amd throttling down when getting hot, but even that will crash if it gets bad enough). install kernel-tools and run turbostat it will show the cpu freqs and cpu temps.
My CPU is Intel. I'll try kernel-tools.
Today is the first time firefox froze on me since leaving turbostat running.
-- Michael hennebry@mail.cs.ndsu.NoDak.edu "His longhand was fairly good in his youth, but as he got older it got smaller, more scribbly, and harder to read; although, like being hanged, one can get used to it." -- Gordon Dickson on H.P. Lovecraft's handwriting -- _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 11/25/2023 12:08 AM, Tim via users wrote:
So, come OS install and update times, I tend to open a box, inspect heatsinks, clean it, and reseat all the connections. Having a spare power supply to swap over is handy, too. They don't always age well, especially the bargain basement types.
Have you ever used a gravity-assist to reseat everything at once?
Tim:
So, come OS install and update times, I tend to open a box, inspect heatsinks, clean it, and reseat all the connections. Having a spare power supply to swap over is handy, too. They don't always age well, especially the bargain basement types.
Joe Zeff:
Have you ever used a gravity-assist to reseat everything at once?
Ha ha, no. But I have seen another guy pick up a boom box and bang it on the table several times until an intermittent went away.
A friend gave me his expensive graphics card that had died. He got pissed off with his PC and punched it and threw it around on the concrete floor. The case was amusingly dented, and a transistor fell off the graphics card. I soldered it back on and it worked fine. It could well have been the cause of all his woes, being badly soldered on in the first place, and intermittently failing.
He'd given me another graphics card before that had overheated when its cooling fan seized, then the heat melted the plastic fan. I just aimed a case mounted fan, it worked fine for many years that way until I retired the system for being ancient and slow compared to current hardware.
On 23 Nov 2023 at 18:53, Michael Hennebry wrote:
Date sent: Thu, 23 Nov 2023 18:53:41 -0600 (CST) From: Michael Hennebry hennebry@web.cs.ndsu.nodak.edu To: noloader@gmail.com, Community support for Fedora users users@lists.fedoraproject.org Subject: Re: firefox keeps freezing on me Send reply to: Community support for Fedora users users@lists.fedoraproject.org
On Thu, 23 Nov 2023, Jeffrey Walton wrote:
On Thu, Nov 23, 2023 at 2:21?PM Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
By now, I am fairly sure that the reason firefox freezes on me is neither a lack of memory nor a lack of CPU. My inference is that it is waiting for something. How do I discover what?
I would investigate the virtual memory system. That's based on your reply to Roger Heflin, and the 8GB of RAM and no swap file. Maybe
Might want to run firefox from a command line and see if it reports anything.
I don't get a freeze, but do see this, perhaps something will be shown?
firefox ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. [Parent 346000, Main Thread] WARNING: Failed to call GetIdletime(): GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.gnome.Mutter.IdleMonitor was not provided by any .service files : 'glib warning', file /builddir/build/BUILD/firefox-120.0/toolkit/xre/nsSigHandlers.cpp:1 87
** (firefox:346000): WARNING **: 11:01:58.425: Failed to call GetIdletime(): GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.gnome.Mutter.IdleMonitor was not provided by any .service files
You mean this: Michael Hennebry wrote (some whitespace deleted):
$ free -h total used free shared buff/cache available Mem: 7.7Gi 4.1Gi 882Mi 572Mi 2.7Gi 2.7Gi Swap: 7.7Gi 98Mi 7.6Gi $ I have no swap partition.
? I'd have thought 882Mi free and 2.7Gi available would be good. BTW I have a lot of tabs, but not a lot of videos. Mostly my tabs are things to read.
start with `vm.overcommit_memory = 2` in `/etc/sysctl.conf`. The 2 says, "say no if we don't have the memory".
From the persistence of the belief that I am running out of memory, I infer one of two scenarios is assumed: 1: firefox is waiting on memory is was told it has, but might never get. 2: firefox has been told memory is unavailable, but does not do anything sensible with that information. The output from free would seem to preclude both scenarios.
What is the state of a process that is waiting on commited, but unavailable, memory?
-- Michael hennebry@mail.cs.ndsu.NoDak.edu "I was just thinking about the life of a pumpkin. Grow up in the sun, happily entwined with others, and then someone comes along, cuts you open, and rips your guts out." -- Buffy -- _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
+------------------------------------------------------------+ Michael D. Setzer II - Computer Science Instructor (Retired) mailto:mikes@guam.net mailto:msetzerii@gmail.com Guam - Where America's Day Begins G4L Disk Imaging Project maintainer http://sourceforge.net/projects/g4l/ +------------------------------------------------------------+
-
Jeffrey Walton -
Joe Zeff -
Michael D. Setzer II -
Michael Hennebry -
richard emberson -
Roger Heflin -
Samuel Sieb -
simmon simmon -
stan -
Tim