cron job question (for checking kernel updates)

[Ranjan Maitra]

Hi,

I have a cron job running which yum updates all my machines once a day.
All of these work fine.

I also have a cron job which checks for kernel updates every hour and
sends me a message if an updated kernel has been installed. Here is the
relevant script:

--- begin file called check-kernel.sh in my scripts/yum directory---

#!/bin/bash

latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')

#echo "$latestkernel"

if uname -a | grep -qv "$latestkernel"; then
notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
-r) but lat est installed rpm is ${latestkernel}; REBOOT required" 
fi;

--- end file called check-kernel.sh ---

(A bit wordy, I know.)

I have verified that this is executable and works from the commandline
without any errors.

I have the following set up via crontab -e:

0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh

The first line downloads cloudmaps for use with xplanet and works just
fine (or appears to) and has done so for aeons.

However, nothing happens (or appears to) for the second line.

What is wrong here?

Many thanks and best wishes,
Ranjan

[Matthew Miller]

On Wed, Oct 02, 2013 at 10:10:45AM -0500, Ranjan Maitra wrote:
> I also have a cron job which checks for kernel updates every hour and
> sends me a message if an updated kernel has been installed. Here is the
> relevant script:

Back when I did the BU Linux distribution, we had a similar script, which
you can still get from

ftp://linst.bu.edu/yum/5.0/source/x86_64/SRPMS/kervercheck-1.0.4-bu50.1.src.rpm

You may find this useful.

I've been meaning to package this up for Fedora but have never gotten around
to it. If someone else would like to, that would be awesome.

This script sends by e-mail, and therefore avoids what I think is the
problem you have: notify-send attaches to the dbus message bus in your
current X session, and jobs run distinct from that.

I haven't tried this, and it's been a long time since I've tried anything
similar, but I think the steps here
http://earlruby.org/2008/08/update-pidgin-status-using-cron/
should help -- it's someone trying to do basically the same thing. (There's
an example for notify-send).

[Rick Stevens]

On 10/02/2013 08:10 AM, Ranjan Maitra issued this missive:
> Hi,
>
> I have a cron job running which yum updates all my machines once a day.
> All of these work fine.
>
> I also have a cron job which checks for kernel updates every hour and
> sends me a message if an updated kernel has been installed. Here is the
> relevant script:
>
> --- begin file called check-kernel.sh in my scripts/yum directory---
>
> #!/bin/bash
>
> latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
>
> #echo "$latestkernel"
>
> if uname -a | grep -qv "$latestkernel"; then
> notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> fi;
>
> --- end file called check-kernel.sh ---
>
> (A bit wordy, I know.)
>
> I have verified that this is executable and works from the commandline
> without any errors.
>
> I have the following set up via crontab -e:
>
> 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
>
> The first line downloads cloudmaps for use with xplanet and works just
> fine (or appears to) and has done so for aeons.
>
> However, nothing happens (or appears to) for the second line.
>
> What is wrong here?

The most common problem I've seen is that cron jobs typically have very
limited paths and the executables you're running (notify-send, uname)
may not be in the path the cronjob has. For giggles, do "echo $PATH" as
the user the cronjob will run as. Then build a cron job that echos
cron's concept of the path:

#!/bin/bash
    echo $PATH >/tmp/pathecho.txt

and have cron run that once as the same user. Look at the data in
/tmp/pathecho.txt and I'll bet you'll find its far more restricted than
that of an interactive shell. To fix it, put in a

PATH="data from the echo $PATH command"

right after the "#!/bin/bash" line in your cron job. Since most of
your commands are in /bin, you could do:

PATH=$PATH:/bin

and that would probably suffice (just making sure /bin is in the cron's
path).
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital    [e-mail] -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
- Grabel's Law: 2 is not equal to 3--not even for large values of 2. -
----------------------------------------------------------------------

[Ranjan Maitra]

On Wed, 2 Oct 2013 09:26:32 -0700 Rick Stevens 
wrote:
> On 10/02/2013 08:10 AM, Ranjan Maitra issued this missive:
> > Hi,
> >
> > I have a cron job running which yum updates all my machines once a day.
> > All of these work fine.
> >
> > I also have a cron job which checks for kernel updates every hour and
> > sends me a message if an updated kernel has been installed. Here is the
> > relevant script:
> >
> > --- begin file called check-kernel.sh in my scripts/yum directory---
> >
> > #!/bin/bash
> >
> > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> >
> > #echo "$latestkernel"
> >
> > if uname -a | grep -qv "$latestkernel"; then
> > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > fi;
> >
> > --- end file called check-kernel.sh ---
> >
> > (A bit wordy, I know.)
> >
> > I have verified that this is executable and works from the commandline
> > without any errors.
> >
> > I have the following set up via crontab -e:
> >
> > 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> > 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
> >
> > The first line downloads cloudmaps for use with xplanet and works just
> > fine (or appears to) and has done so for aeons.
> >
> > However, nothing happens (or appears to) for the second line.
> >
> > What is wrong here?
>
> The most common problem I've seen is that cron jobs typically have very
> limited paths and the executables you're running (notify-send, uname)
> may not be in the path the cronjob has. For giggles, do "echo $PATH" as
> the user the cronjob will run as. Then build a cron job that echos
> cron's concept of the path:
>
> #!/bin/bash
>    echo $PATH >/tmp/pathecho.txt
>
> and have cron run that once as the same user. Look at the data in
> /tmp/pathecho.txt and I'll bet you'll find its far more restricted than
> that of an interactive shell.

I tried this: here is what I got --

$ echo $PATH
/bin:/usr/bin:/usr/local/bin:/usr/local/sbin:/usr/sbin:/home/user/bin

$ cat /tmp/pathecho.txt 
/usr/bin:/bin

Clearly more restrictive as you suggest.

However, what I do not understand is why this does not "work" though. I
do not understand because notify-send is in /usr/bin as well as in /bin:

$ which notify-send 
notify-send is /bin/notify-send
notify-send is /usr/bin/notify-send

Both appear to be in the path. So where is the problem?

I will also try the suggestions in the references in Matthew Miller's
response now.

Many thanks,
Ranjan
> To fix it, put in a
>
> PATH="data from the echo $PATH command"
>
> right after the "#!/bin/bash" line in your cron job. Since most of
> your commands are in /bin, you could do:
>
> PATH=$PATH:/bin
>
> and that would probably suffice (just making sure /bin is in the cron's
> path).

[Rick Stevens]

On 10/02/2013 01:06 PM, Ranjan Maitra issued this missive:
> On Wed, 2 Oct 2013 09:26:32 -0700 Rick Stevens 
> wrote:
> > On 10/02/2013 08:10 AM, Ranjan Maitra issued this missive:
> > > Hi,
> > >
> > > I have a cron job running which yum updates all my machines once a day.
> > > All of these work fine.
> > >
> > > I also have a cron job which checks for kernel updates every hour and
> > > sends me a message if an updated kernel has been installed. Here is the
> > > relevant script:
> > >
> > > --- begin file called check-kernel.sh in my scripts/yum directory---
> > >
> > > #!/bin/bash
> > >
> > > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> > >
> > > #echo "$latestkernel"
> > >
> > > if uname -a | grep -qv "$latestkernel"; then
> > > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > > fi;
> > >
> > > --- end file called check-kernel.sh ---
> > >
> > > (A bit wordy, I know.)
> > >
> > > I have verified that this is executable and works from the commandline
> > > without any errors.
> > >
> > > I have the following set up via crontab -e:
> > >
> > > 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> > > 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
> > >
> > > The first line downloads cloudmaps for use with xplanet and works just
> > > fine (or appears to) and has done so for aeons.
> > >
> > > However, nothing happens (or appears to) for the second line.
> > >
> > > What is wrong here?
> >
> > The most common problem I've seen is that cron jobs typically have very
> > limited paths and the executables you're running (notify-send, uname)
> > may not be in the path the cronjob has. For giggles, do "echo $PATH" as
> > the user the cronjob will run as. Then build a cron job that echos
> > cron's concept of the path:
> >
> > #!/bin/bash
> >    echo $PATH >/tmp/pathecho.txt
> >
> > and have cron run that once as the same user. Look at the data in
> > /tmp/pathecho.txt and I'll bet you'll find its far more restricted than
> > that of an interactive shell.
>
> I tried this: here is what I got --
>
> $ echo $PATH
> /bin:/usr/bin:/usr/local/bin:/usr/local/sbin:/usr/sbin:/home/user/bin
>
> $ cat /tmp/pathecho.txt
> /usr/bin:/bin
>
> Clearly more restrictive as you suggest.
>
> However, what I do not understand is why this does not "work" though. I
> do not understand because notify-send is in /usr/bin as well as in /bin:
>
> $ which notify-send
> notify-send is /bin/notify-send
> notify-send is /usr/bin/notify-send
>
> Both appear to be in the path. So where is the problem?
>
> I will also try the suggestions in the references in Matthew Miller's
> response now.

You could try adding a "set -x" after the shebang line:

#!/bin/bash
    set -x

That tells bash to print out every expansion. Then change your cron
entry to have a ">/tmp/test.txt 2>&1" at the end:

5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh >/tmp/test.txt 2>&1

Let the script run, then examine the /tmp/test.txt file to see what bash
doesn't like. Don't forget to remove the "set -x" from the script and
the ">... 2>&1" stuff from the crontab entry when you're done. Odds are
it's a path issue or something that isn't expanding as you think it
would because it's not an interactive shell running the script.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital    [e-mail] -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
-                He who laughs last thinks slowest.                  -
----------------------------------------------------------------------

[Ranjan Maitra]

On Wed, 2 Oct 2013 13:34:36 -0700 Rick Stevens 
wrote:
> On 10/02/2013 01:06 PM, Ranjan Maitra issued this missive:
> > On Wed, 2 Oct 2013 09:26:32 -0700 Rick Stevens 
> > wrote:
> > > On 10/02/2013 08:10 AM, Ranjan Maitra issued this missive:
> > > > Hi,
> > > >
> > > > I have a cron job running which yum updates all my machines once a day.
> > > > All of these work fine.
> > > >
> > > > I also have a cron job which checks for kernel updates every hour and
> > > > sends me a message if an updated kernel has been installed. Here is the
> > > > relevant script:
> > > >
> > > > --- begin file called check-kernel.sh in my scripts/yum directory---
> > > >
> > > > #!/bin/bash
> > > >
> > > > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> > > >
> > > > #echo "$latestkernel"
> > > >
> > > > if uname -a | grep -qv "$latestkernel"; then
> > > > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > > > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > > > fi;
> > > >
> > > > --- end file called check-kernel.sh ---
> > > >
> > > > (A bit wordy, I know.)
> > > >
> > > > I have verified that this is executable and works from the commandline
> > > > without any errors.
> > > >
> > > > I have the following set up via crontab -e:
> > > >
> > > > 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> > > > 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
> > > >
> > > > The first line downloads cloudmaps for use with xplanet and works just
> > > > fine (or appears to) and has done so for aeons.
> > > >
> > > > However, nothing happens (or appears to) for the second line.
> > > >
> > > > What is wrong here?
> > >
> > > The most common problem I've seen is that cron jobs typically have very
> > > limited paths and the executables you're running (notify-send, uname)
> > > may not be in the path the cronjob has. For giggles, do "echo $PATH" as
> > > the user the cronjob will run as. Then build a cron job that echos
> > > cron's concept of the path:
> > >
> > > #!/bin/bash
> > >    echo $PATH >/tmp/pathecho.txt
> > >
> > > and have cron run that once as the same user. Look at the data in
> > > /tmp/pathecho.txt and I'll bet you'll find its far more restricted than
> > > that of an interactive shell.
> >
> > I tried this: here is what I got --
> >
> > $ echo $PATH
> > /bin:/usr/bin:/usr/local/bin:/usr/local/sbin:/usr/sbin:/home/user/bin
> >
> > $ cat /tmp/pathecho.txt
> > /usr/bin:/bin
> >
> > Clearly more restrictive as you suggest.
> >
> > However, what I do not understand is why this does not "work" though. I
> > do not understand because notify-send is in /usr/bin as well as in /bin:
> >
> > $ which notify-send
> > notify-send is /bin/notify-send
> > notify-send is /usr/bin/notify-send
> >
> > Both appear to be in the path. So where is the problem?
> >
> > I will also try the suggestions in the references in Matthew Miller's
> > response now.
>
> You could try adding a "set -x" after the shebang line:
>
> #!/bin/bash
>    set -x
>
> That tells bash to print out every expansion. Then change your cron
> entry to have a ">/tmp/test.txt 2>&1" at the end:
>
> 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh >/tmp/test.txt 2>&1
>
> Let the script run, then examine the /tmp/test.txt file to see what bash
> doesn't like. Don't forget to remove the "set -x" from the script and
> the ">... 2>&1" stuff from the crontab entry when you're done. Odds are
> it's a path issue or something that isn't expanding as you think it
> would because it's not an interactive shell running the script.

Here is what I get:

$ more /tmp/test.txt 
++ tail -n1
++ rpm -q kernel
++ sed -e s/kernel-//
+ latestkernel=3.11.2-201.fc19.i686
+ grep -qv 3.11.2-201.fc19.i686
+ uname -a
++ uname -r
+ notify-send 'Kernel UPDATE on subarnarekha.stat.iastate.edu: Running
Kernel is 3.11.1-200.fc19.i686 but latest installed rpm is
3.11.2-201.fc19.i686; REBOOT r equired'

I also tried explicitly putting in /usr/bin/notify-send (hardcoding it
in) to the original code, but no effect.

THanks,
Ranjan

____________________________________________________________
FREE ONLINE PHOTOSHARING - Share your photos online with your friends and family!
Visit http://www.inbox.com/photosharing to find out more!

[Rick Stevens]

On 10/02/2013 02:27 PM, Ranjan Maitra issued this missive:
> On Wed, 2 Oct 2013 13:34:36 -0700 Rick Stevens 
> wrote:
> > On 10/02/2013 01:06 PM, Ranjan Maitra issued this missive:
> > > On Wed, 2 Oct 2013 09:26:32 -0700 Rick Stevens 
> > > wrote:
> > > > On 10/02/2013 08:10 AM, Ranjan Maitra issued this missive:
> > > > > Hi,
> > > > >
> > > > > I have a cron job running which yum updates all my machines once a day.
> > > > > All of these work fine.
> > > > >
> > > > > I also have a cron job which checks for kernel updates every hour and
> > > > > sends me a message if an updated kernel has been installed. Here is the
> > > > > relevant script:
> > > > >
> > > > > --- begin file called check-kernel.sh in my scripts/yum directory---
> > > > >
> > > > > #!/bin/bash
> > > > >
> > > > > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> > > > >
> > > > > #echo "$latestkernel"
> > > > >
> > > > > if uname -a | grep -qv "$latestkernel"; then
> > > > > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > > > > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > > > > fi;
> > > > >
> > > > > --- end file called check-kernel.sh ---
> > > > >
> > > > > (A bit wordy, I know.)
> > > > >
> > > > > I have verified that this is executable and works from the commandline
> > > > > without any errors.
> > > > >
> > > > > I have the following set up via crontab -e:
> > > > >
> > > > > 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> > > > > 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
> > > > >
> > > > > The first line downloads cloudmaps for use with xplanet and works just
> > > > > fine (or appears to) and has done so for aeons.
> > > > >
> > > > > However, nothing happens (or appears to) for the second line.
> > > > >
> > > > > What is wrong here?
> > > >
> > > > The most common problem I've seen is that cron jobs typically have very
> > > > limited paths and the executables you're running (notify-send, uname)
> > > > may not be in the path the cronjob has. For giggles, do "echo $PATH" as
> > > > the user the cronjob will run as. Then build a cron job that echos
> > > > cron's concept of the path:
> > > >
> > > > #!/bin/bash
> > > >    echo $PATH >/tmp/pathecho.txt
> > > >
> > > > and have cron run that once as the same user. Look at the data in
> > > > /tmp/pathecho.txt and I'll bet you'll find its far more restricted than
> > > > that of an interactive shell.
> > >
> > > I tried this: here is what I got --
> > >
> > > $ echo $PATH
> > > /bin:/usr/bin:/usr/local/bin:/usr/local/sbin:/usr/sbin:/home/user/bin
> > >
> > > $ cat /tmp/pathecho.txt
> > > /usr/bin:/bin
> > >
> > > Clearly more restrictive as you suggest.
> > >
> > > However, what I do not understand is why this does not "work" though. I
> > > do not understand because notify-send is in /usr/bin as well as in /bin:
> > >
> > > $ which notify-send
> > > notify-send is /bin/notify-send
> > > notify-send is /usr/bin/notify-send
> > >
> > > Both appear to be in the path. So where is the problem?
> > >
> > > I will also try the suggestions in the references in Matthew Miller's
> > > response now.
> >
> > You could try adding a "set -x" after the shebang line:
> >
> > #!/bin/bash
> >    set -x
> >
> > That tells bash to print out every expansion. Then change your cron
> > entry to have a ">/tmp/test.txt 2>&1" at the end:
> >
> > 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh >/tmp/test.txt 2>&1
> >
> > Let the script run, then examine the /tmp/test.txt file to see what bash
> > doesn't like. Don't forget to remove the "set -x" from the script and
> > the ">... 2>&1" stuff from the crontab entry when you're done. Odds are
> > it's a path issue or something that isn't expanding as you think it
> > would because it's not an interactive shell running the script.
>
> Here is what I get:
>
> $ more /tmp/test.txt
> ++ tail -n1
> ++ rpm -q kernel
> ++ sed -e s/kernel-//
> latestkernel=3.11.2-201.fc19.i686
> grep -qv 3.11.2-201.fc19.i686
> uname -a
>
> ++ uname -r
> notify-send 'Kernel UPDATE on subarnarekha.stat.iastate.edu: Running
>
> Kernel is 3.11.1-200.fc19.i686 but latest installed rpm is
> 3.11.2-201.fc19.i686; REBOOT r equired'
>
> I also tried explicitly putting in /usr/bin/notify-send (hardcoding it
> in) to the original code, but no effect.

Ok, so it looks like there's nothing wrong in the script so you can 
remove the "set -x" from the script and the ">/tmp/test.txt 2>&1" from
the crontab.

My guess is that the cron doesn't know that there is a GUI running and
so notify-send doesn't know where to send the message. I've seen some
stuff regarding this in one of the Ubuntu forums. Have a look at this:

http://askubuntu.com/questions/298608/notify-send-doesnt-work-from-crontab

In particular, look at the second and third answers (from danatela and 
krlmlr) and combine them (from danatela's answer to find who's logged
in on the GUI and krlmlr's answer to set the DBUS_SESSION_BUS_ADDRESS
for that user).
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital    [e-mail] -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
-           "I understand Windows 2000 has a Y2K problem."           -
----------------------------------------------------------------------

[poma]

On 02.10.2013 22:06, Ranjan Maitra wrote:
…
> However, what I do not understand is why this does not "work" though. I
> do not understand because notify-send is in /usr/bin as well as in /bin:
>
> $ which notify-send 
> notify-send is /bin/notify-send
> notify-send is /usr/bin/notify-send
>
> Both appear to be in the path. So where is the problem?

http://goo.gl/Mqq6UP

poma

[Ranjan Maitra]

On Thu, 3 Oct 2013 02:17:09 +0200 poma 
wrote:
> On 02.10.2013 22:06, Ranjan Maitra wrote:
> …
> > However, what I do not understand is why this does not "work" though. I
> > do not understand because notify-send is in /usr/bin as well as in /bin:
> >
> > $ which notify-send 
> > notify-send is /bin/notify-send
> > notify-send is /usr/bin/notify-send
> >
> > Both appear to be in the path. So where is the problem?
>
> http://goo.gl/Mqq6UP

Thanks! But this does not "work" either. Here is what I tried in my
crontab -e:

0 * * * * maitra DISPLAY=:0.0 $HOME/scripts/yum/check-kernel.sh

(maitra is my username).

Here is how I changed my check-kernel.sh, though I am not sure if there
was a problem with the earlier one....

$more check-kernel.sh

#!/bin/bash
# from http://goo.gl/Mqq6UP

latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')

#echo "$latestkernel"

if uname -a | grep -qv "$latestkernel"; then
notify-send "Kernel UPDATE on ${HOSTNAME}: " \
"<b><i>Running</i></b> kernel is 
<b><i>$(uname -r | awk '{gsub(/*PAE/,""); print}')</i></b>
but <b>latest</b> <b><u>installed</u></b> is
<b><u>${latestkernel}</u></b>
required
<b><u>REBOOT</u></b>"
fi;

How do you get yours to work?

(Works fine from the commandline, of course.) Btw, FWIW, I am using
dunst.

Many thanks,
Ranjan

____________________________________________________________
FREE 3D MARINE AQUARIUM SCREENSAVER - Watch dolphins, sharks & orcas on your desktop!
Check it out at http://www.inbox.com/marineaquarium

[Matthew Miller]

On Wed, Oct 02, 2013 at 07:55:51PM -0500, Ranjan Maitra wrote:
> 0 * * * * maitra DISPLAY=:0.0 $HOME/scripts/yum/check-kernel.sh

Did you try the things in the blog post I'd linked to? As I said in my
previous method, notify-send needs to talk to your dbus session, and it
can't get that from inside cron.

http://earlruby.org/2008/08/update-pidgin-status-using-cron/

has a way of working around that.

[Ranjan Maitra]

On Wed, 2 Oct 2013 21:38:26 -0400 Matthew Miller
 wrote:
> On Wed, Oct 02, 2013 at 07:55:51PM -0500, Ranjan Maitra wrote:
> > 0 * * * * maitra DISPLAY=:0.0 $HOME/scripts/yum/check-kernel.sh
>
> Did you try the things in the blog post I'd linked to? As I said in my
> previous method, notify-send needs to talk to your dbus session, and it
> can't get that from inside cron.
>
> http://earlruby.org/2008/08/update-pidgin-status-using-cron/
>
> has a way of working around that.

Sorry, I was going to get to this, but wanted to try the other
quicker-implemented suggestions also. Now that there is a sort of
workaround there, I will look into implementing this. (This looks more
elegant to me: maybe it is no different from the other solution.)

Thanks again!
Ranjan
> -- 
> Matthew Miller  ☁☁☁  Fedora Cloud Architect  ☁☁☁  
> -- 
> users mailing list
> [e-mail]
> To unsubscribe or change subscription options:
> https://admin.fedoraproject.org/mailman/listinfo/users
> Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
> Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
> Have a question? Ask away: http://ask.fedoraproject.org

(SOLVED) Re: cron job question (for checking kernel updates)

[Ranjan Maitra]

On Wed, 2 Oct 2013 21:38:26 -0400 Matthew Miller
 wrote:
> On Wed, Oct 02, 2013 at 07:55:51PM -0500, Ranjan Maitra wrote:
> > 0 * * * * maitra DISPLAY=:0.0 $HOME/scripts/yum/check-kernel.sh
>
> Did you try the things in the blog post I'd linked to? As I said in my
> previous method, notify-send needs to talk to your dbus session, and it
> can't get that from inside cron.
>
> http://earlruby.org/2008/08/update-pidgin-status-using-cron/
>
> has a way of working around that.

Following, this post pointed to me my Matthew Miller, I am able to get
things working. Here is what I did (same as in the post there).

Created a file as follows (called ~/bin/export_x_info):

#!/bin/bash
# Export the dbus session address on startup so it can be used by cron
touch $HOME/.Xdbus
chmod 600 $HOME/.Xdbus
env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
# Export XAUTHORITY value on startup so it can be used by cron
env | grep XAUTHORITY >> $HOME/.Xdbus
echo 'export XAUTHORITY' >> $HOME/.Xdbus

----
make the file executable:

$ chmod 700 ~/bin/export_x_info

Change this to be sourced at startup. I added the line to
~/.config/openbox/autostart:

/home/maitra/bin/export_x_info&

----

added the following line to crontab (via crontab -e):

37 * * * * source $HOME/.Xdbus; nice -n 19
$HOME/scripts/yum/check-kernel.sh

This works now!

Thanks very much everybody's help!
Best wishes,
Ranjan

____________________________________________________________
FREE 3D MARINE AQUARIUM SCREENSAVER - Watch dolphins, sharks & orcas on your desktop!
Check it out at http://www.inbox.com/marineaquarium

(SOLVED) Re: cron job question (for checking kernel updates)

[Joe Zeff]

On 10/03/2013 02:38 PM, Ranjan Maitra wrote:
> #!/bin/bash
> # Export the dbus session address on startup so it can be used by cron
> touch $HOME/.Xdbus
> chmod 600 $HOME/.Xdbus
> env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
> echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
> # Export XAUTHORITY value on startup so it can be used by cron
> env | grep XAUTHORITY >> $HOME/.Xdbus
> echo 'export XAUTHORITY' >> $HOME/.Xdbus

The first executable line is redundant, because when you redirect the 
output of the env to ~/.Xdbus, that will create it, meaning that the 
chmod line can be moved down one, so that the file exists when the 
command is executed.  I'm not saying that you have to do it this way, 
but it just looks cleaner the other way, at least to me.  YMMV, and it's 
your system.

(SOLVED) Re: cron job question (for checking kernel updates)

[Ranjan Maitra]

On Thu, 3 Oct 2013 15:00:55 -0700 Joe Zeff  wrote:
> On 10/03/2013 02:38 PM, Ranjan Maitra wrote:
> > #!/bin/bash
> > # Export the dbus session address on startup so it can be used by cron
> > touch $HOME/.Xdbus
> > chmod 600 $HOME/.Xdbus
> > env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
> > echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
> > # Export XAUTHORITY value on startup so it can be used by cron
> > env | grep XAUTHORITY >> $HOME/.Xdbus
> > echo 'export XAUTHORITY' >> $HOME/.Xdbus
>
> The first executable line is redundant, because when you redirect the 
> output of the env to ~/.Xdbus, that will create it, meaning that the 
> chmod line can be moved down one, so that the file exists when the 
> command is executed.  I'm not saying that you have to do it this way, 
> but it just looks cleaner the other way, at least to me.  YMMV, and it's 
> your system.

I think I agree with you. So, basically, then have the following in
~/bin/export_x_info:

#!/bin/bash
# Export the dbus session address on startup so it can be used by cron
env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
chmod 600 $HOME/.Xdbus
echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
# Export XAUTHORITY value on startup so it can be used by cron
env | grep XAUTHORITY >> $HOME/.Xdbus
echo 'export XAUTHORITY' >> $HOME/.Xdbus

Thanks,
Ranjan

(SOLVED) Re: cron job question (for checking kernel updates)

[Matthew Miller]

On Thu, Oct 03, 2013 at 03:00:55PM -0700, Joe Zeff wrote:
> > touch $HOME/.Xdbus
> > chmod 600 $HOME/.Xdbus
> > env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
> > echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
> > # Export XAUTHORITY value on startup so it can be used by cron
> > env | grep XAUTHORITY >> $HOME/.Xdbus
> > echo 'export XAUTHORITY' >> $HOME/.Xdbus
>
> The first executable line is redundant, because when you redirect
> the output of the env to ~/.Xdbus, that will create it, meaning that
> the chmod line can be moved down one, so that the file exists when
> the command is executed.  I'm not saying that you have to do it this
> way, but it just looks cleaner the other way, at least to me.  YMMV,
> and it's your system.

It's not a big deal, but if you do it that way, there will be a brief
instant when the permissions may be more open than 600, during which someone
else could maybe steal the contents. On Fedora, the default will be 664,
which is world-readable. Your home directory may also be readable by others,
so they could theeeeretically try to steal your DBUS_SESSION_BUS_ADDRESS.

(SOLVED) Re: cron job question (for checking kernel updates)

[poma]

On 03.10.2013 23:38, Ranjan Maitra wrote:
> Following, this post pointed to me my Matthew Miller, I am able to get
> things working. Here is what I did (same as in the post there).
>
> Created a file as follows (called ~/bin/export_x_info):
>
> #!/bin/bash
> # Export the dbus session address on startup so it can be used by cron
> touch $HOME/.Xdbus
> chmod 600 $HOME/.Xdbus
> env | grep DBUS_SESSION_BUS_ADDRESS > $HOME/.Xdbus
> echo 'export DBUS_SESSION_BUS_ADDRESS' >> $HOME/.Xdbus
> # Export XAUTHORITY value on startup so it can be used by cron
> env | grep XAUTHORITY >> $HOME/.Xdbus
> echo 'export XAUTHORITY' >> $HOME/.Xdbus
>
> make the file executable:
>
> $ chmod 700 ~/bin/export_x_info
>
> Change this to be sourced at startup. I added the line to
> ~/.config/openbox/autostart:
>
> /home/maitra/bin/export_x_info&
>
> added the following line to crontab (via crontab -e):
>
> 37 * * * * source $HOME/.Xdbus; nice -n 19
> $HOME/scripts/yum/check-kernel.sh
>
> This works now!
>
> Thanks very much everybody's help!
> Best wishes,
> Ranjan

For this we use The Desktop Environments and The Display Managers. :)
For the full GNU/Linux experience, I recommend something called LFS[1],
Now, that's a real fun. ;)
Remember, the second
you go through those doors,
everything changes.
Our old life is done.
Let's go for a little ride.

poma

[1] http://en.wikipedia.org/wiki/Linux_From_Scratch

[Joe Zeff]

On 10/02/2013 05:55 PM, Ranjan Maitra wrote:
> Thanks! But this does not "work" either.

There may be a much easier way: before your cron job starts the update, 
have it do this:

yum check-update > /tmp/updates.txt

Then, after the update, grep /tmp/updates.txt looking for kernel, 
redirecting the output to /dev/null because all you need is the exit 
status.  If there are any matches, it returns 0 and the script can 
notify you that it needs rebooting. (Sending email might be the best 
way.)  If not, it does nothing.  And, the nice thing about this is that 
it ends up being exactly one script, run once a day.

[poma]

On 03.10.2013 02:55, Ranjan Maitra wrote:
> Thanks! But this does not "work" either. Here is what I tried in my
> crontab -e:
>
> 0 * * * * maitra DISPLAY=:0.0 $HOME/scripts/yum/check-kernel.sh
>
> (maitra is my username).
>
> Here is how I changed my check-kernel.sh, though I am not sure if there
> was a problem with the earlier one....
>
> $more check-kernel.sh
>
> #!/bin/bash
> # from http://goo.gl/Mqq6UP
>
> latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
>
> #echo "$latestkernel"
>
> if uname -a | grep -qv "$latestkernel"; then
> notify-send "Kernel UPDATE on ${HOSTNAME}: " \
> "<b><i>Running</i></b> kernel is 
> <b><i>$(uname -r | awk '{gsub(/*PAE/,""); print}')</i></b>
> but <b>latest</b> <b><u>installed</u></b> is
> <b><u>${latestkernel}</u></b>
> required
> <b><u>REBOOT</u></b>"
> fi;
>
> How do you get yours to work?
>
> (Works fine from the commandline, of course.) Btw, FWIW, I am using
> dunst.

http://goo.gl/Mqq6UP
Cronie-notify-send-Dunst-Openbox.webm

poma

[Joe Zeff]

On 10/02/2013 08:10 AM, Ranjan Maitra wrote:
> I also have a cron job which checks for kernel updates every hour and
> sends me a message if an updated kernel has been installed.

Why?  If you only update the machine once a day, checking for a new 
kernel 24 times per day is redundant.  Either add it to the script 
running yum, or have it run once a day, an hour after the update.

[inode0]

On Wed, Oct 2, 2013 at 10:10 AM, Ranjan Maitra
 wrote:
> Hi,
>
> I have a cron job running which yum updates all my machines once a day.
> All of these work fine.
>
> I also have a cron job which checks for kernel updates every hour and
> sends me a message if an updated kernel has been installed. Here is the
> relevant script:
>
> --- begin file called check-kernel.sh in my scripts/yum directory---
>
> #!/bin/bash

export DISPLAY=:0
> latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
>
> #echo "$latestkernel"
>
> if uname -a | grep -qv "$latestkernel"; then
> notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> fi;
>
> --- end file called check-kernel.sh ---

Give that a try if you haven't yet Ranjan.

John

[Ranjan Maitra]

On Wed, 2 Oct 2013 20:53:52 -0500 inode0  wrote:
> On Wed, Oct 2, 2013 at 10:10 AM, Ranjan Maitra
>  wrote:
> > Hi,
> >
> > I have a cron job running which yum updates all my machines once a day.
> > All of these work fine.
> >
> > I also have a cron job which checks for kernel updates every hour and
> > sends me a message if an updated kernel has been installed. Here is the
> > relevant script:
> >
> > --- begin file called check-kernel.sh in my scripts/yum directory---
> >
> > #!/bin/bash
>
> export DISPLAY=:0
> > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> >
> > #echo "$latestkernel"
> >
> > if uname -a | grep -qv "$latestkernel"; then
> > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > fi;
> >
> > --- end file called check-kernel.sh ---

Got it! This works. However, I will also look at the link suggested by
Matthew Miller (and also by Rick later on). I suspect that that is more
general.

Many thanks again for all the help! This is what makes this mailing
list the best!! The link suggested by Matthew Miller
(  http://earlruby.org/2008/08/update-pidgin-status-using-cron/ )
and the points provided by all others provided a lot of explanation and
shed light into things I did not know much about.

Ranjan

____________________________________________________________
GET FREE 5GB EMAIL - Check out spam free email with many cool features!
Visit http://www.inbox.com/email to find out more!

[Ranjan Maitra]

On Wed, 2 Oct 2013 20:53:52 -0500 inode0  wrote:
> On Wed, Oct 2, 2013 at 10:10 AM, Ranjan Maitra
>  wrote:
> > Hi,
> >
> > I have a cron job running which yum updates all my machines once a day.
> > All of these work fine.
> >
> > I also have a cron job which checks for kernel updates every hour and
> > sends me a message if an updated kernel has been installed. Here is the
> > relevant script:
> >
> > --- begin file called check-kernel.sh in my scripts/yum directory---
> >
> > #!/bin/bash
>
> export DISPLAY=:0

I was wondering -- why, instead of this, the following would not work:

export $(env | grep DISPLAY)

Note that:

echo  $(env | grep DISPLAY)>/tmp/tmp.txt

creates a blank file so there is a problem, clearly.

Just trying to see if I can easily hardcode the DISPLAY variable.

Thanks!
Ranjan
> > latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
> >
> > #echo "$latestkernel"
> >
> > if uname -a | grep -qv "$latestkernel"; then
> > notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> > -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> > fi;
> >
> > --- end file called check-kernel.sh ---
>
> Give that a try if you haven't yet Ranjan.
> John
>
> users mailing list
> [e-mail]
> To unsubscribe or change subscription options:
> https://admin.fedoraproject.org/mailman/listinfo/users
> Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
> Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
> Have a question? Ask away: http://ask.fedoraproject.org

[inode0]

On Wed, Oct 2, 2013 at 10:42 PM, Ranjan Maitra
 wrote:
> On Wed, 2 Oct 2013 20:53:52 -0500 inode0  wrote:
> > On Wed, Oct 2, 2013 at 10:10 AM, Ranjan Maitra
> >  wrote:
> > > Hi,
> > >
> > > I have a cron job running which yum updates all my machines once a day.
> > > All of these work fine.
> > >
> > > I also have a cron job which checks for kernel updates every hour and
> > > sends me a message if an updated kernel has been installed. Here is the
> > > relevant script:
> > >
> > > --- begin file called check-kernel.sh in my scripts/yum directory---
> > >
> > > #!/bin/bash
> >
> > export DISPLAY=:0
>
> I was wondering -- why, instead of this, the following would not work:
>
> export $(env | grep DISPLAY)

DISPLAY is not set in the environment used by cron. That was the whole
problem to begin with.

John

[Bjørge Solli]

Is the script executable?

-b

On 10/02/2013 05:10 PM, Ranjan Maitra wrote:
> Hi,
>
> I have a cron job running which yum updates all my machines once a day.
> All of these work fine.
>
> I also have a cron job which checks for kernel updates every hour and
> sends me a message if an updated kernel has been installed. Here is the
> relevant script:
>
> --- begin file called check-kernel.sh in my scripts/yum directory---
>
> #!/bin/bash
>
> latestkernel=$(rpm -q kernel |tail -n1|sed -e 's/kernel-//')
>
> #echo "$latestkernel"
>
> if uname -a | grep -qv "$latestkernel"; then
> notify-send "Kernel UPDATE on ${HOSTNAME}: Running Kernel is $(uname
> -r) but lat est installed rpm is ${latestkernel}; REBOOT required"
> fi;
>
> --- end file called check-kernel.sh ---
>
> (A bit wordy, I know.)
>
> I have verified that this is executable and works from the commandline
> without any errors.
>
> I have the following set up via crontab -e:
>
> 0,15,30,45 * * * * nice -n 19 $HOME/.xplanet/download_xplanet_cloudmap
> 5 * * * * nice -n 19 $HOME/scripts/yum/check-kernel.sh
>
> The first line downloads cloudmaps for use with xplanet and works just
> fine (or appears to) and has done so for aeons.
>
> However, nothing happens (or appears to) for the second line.
>
> What is wrong here?
>
> Many thanks and best wishes,
> Ranjan