Topic: Boot Directory Full

[billi]

A quick note to say that on a yum update in 7.3 to 7.4 I got a message that it could not complete because the /boot directory was full.
I tried deleting all but the last 2 sets of the following file groups
inintrd-##
system.map ##
vmlinuz ##
The partitons disk space went from 96% to 36% full
I ran yum update, and all seems well.

[David Harper]

This probably needs to be reported to the bug tracker, if only for reference purposes.

[cactus]

This probably needs to be reported to the bug tracker, if only for reference purposes.

This should already be fixed according to http://bugs.contribs.org/show_bug.cgi?id=2101 .

I think you can work around it like this:

Code: [Select]

yum update smeserver-yum
signal-event post-upgrade
signal-event reboot
When it comes back up wait a few minutes (for the update check to finish) and retry updating then, it should decide to remove kernels that are obsolete freeing up disk space on the /boot.

If that does not work (as it might try install kernels before removing them) you can always uninstall some of the older kernels, using either http://bugs.contribs.org/show_bug.cgi?id=2101#c44 or this http://bugs.contribs.org/show_bug.cgi?id=2101#c36

[billi]

First - Thanks for your help.
Aron Glover wrote in

http://bugs.contribs.org/show_bug.cgi?id=2101#c57

> No. Don't use yum to remove packages. Use rpm. "yum remove xxx" has been known
> to be overenthusiastic.
>
> rpm -e kernel-2.6.9-22.0.1.EL.i686 kernel-smp-2.6.9-22.0.1.EL.i686 ...

I have a backup of the root files of /boot from before the successful removal of the older kernel files (using rm *55* or what ever old file number are)    and a successful "yum update". My question is - Should I put the removed files back and do what was suggested above, or just leave well enough alone.

[cactus]

First - Thanks for your help.
Aron Glover wrote in

http://bugs.contribs.org/show_bug.cgi?id=2101#c57

> No. Don't use yum to remove packages. Use rpm. "yum remove xxx" has been known
> to be overenthusiastic.
>
> rpm -e kernel-2.6.9-22.0.1.EL.i686 kernel-smp-2.6.9-22.0.1.EL.i686 ...

I have a backup of the root files of /boot from before the successful removal of the older kernel files (using rm *55* or what ever old file number are)    and a successful "yum update". My question is - Should I put the removed files back and do what was suggested above, or just leave well enough alone.

It is not completely clear from the commands what you did. Did you execute any of the instructions posted by me or instructions from the two comments mentioned in my previous post?

If not try to describe what you did and why as the solution was in the links I provided. Please next time try to follow-up the advice given to you, this makes it much easier for us to help you.

[billi]

OK - from the start
Tried the yum update - got the /boot full error
Search and got and tried the Bugzilla  fix -
yum update smeserver-yum
signal-event post-upgrade
signal-event reboot

tried the update again
boot disk  full errors again

There were  a few different fixes suggested in the  Bugzilla thread. I could not decide which one to run.
I made a backup of the \boot root files.
I then deleted all be the 2 most recent sets of files using the rm command - thinking - at least giving the update room to run.
Ran the yum update again and it worked.

Still unsure of the what problems might occur if I did not do the RPM based removals -  I posted to this forum for some advice - which thankfully I am getting.

So I guess my question still stands, do you think that I should replace the deleted files from my backup and run and one of the RPM based fixes (not yet clear which one)  or just leave well einough alone.
Thanks

[cactus]

There were  a few different fixes suggested in the  Bugzilla thread. I could not decide which one to run.

That is why I direclty pointed you to two, from which you should have chosen one.

I made a backup of the \boot root files.
I then deleted all be the 2 most recent sets of files using the rm command - thinking - at least giving the update room to run.

And that is what you should not have done as that might leave your system in an unknown, unbootable state. If you were to delete anything you should have been better of removing the oldest ones.

Ran the yum update again and it worked.

You are lucky.

Still unsure of the what problems might occur if I did not do the RPM based removals -  I posted to this forum for some advice - which thankfully I am getting.

So I guess my question still stands, do you think that I should replace the deleted files from my backup and run and one of the RPM based fixes (not yet clear which one)  or just leave well einough alone.
Thanks

You should just have done what was suggested to you:

If that does not work (as it might try install kernels before removing them) you can always uninstall some of the older kernels, using either http://bugs.contribs.org/show_bug.cgi?id=2101#c44 or this http://bugs.contribs.org/show_bug.cgi?id=2101#c36

Please consider that we are trying to help as good as we can, so please do as we suggest instead of just wandering of on your own, while you asked for help and advice. I consider this wasting my valuable and precious time I dedicate to SME Server and to you, so please keep that in mind.

[billi]

In order again.
I did leave the 2 most recent kernel file sets  when I did the rm on the files and did the yum update with them in place. Sorry if that was not clear.

The first few steps were done before my first post here, so I your advice could have only have been considered after the deletion of the files and running of the yum update.

You in fact did suggest 2 short scripts but I don't have your depth of knowledge to make an informed decision on which one to use or if I can try both if one does not work. I was simply trying to ask advice on:
1. should I put the files back so I can run your suggested scripts
2. which of the scripts would be best to try.

I think your patience is growing thin with this topic and me so .......
Thanks again for your efforts, and unless I see otherwise through this thread that it would be best to take another action, I will just leave things the way they are and hope for the best.

[cactus]

In order again.
You in fact did suggest 2 short scripts but I don't have your depth of knowledge to make an informed decision on which one to use or if I can try both if one does not work. I was simply trying to ask advice on:
1. should I put the files back so I can run your suggested scripts
2. which of the scripts would be best to try.

I suggest you run this version of the command to get everything back in sync:

Code: [Select]

rpm -e $(rpm -q kernel kernel-smp kernel-hugemem \
 | sed /$(uname -r | sed -e s/smp//)/d)It safes you the restore and only keeps the current kernel you are running. After that and the latest upgrades you should then never be bothered by this.

[billi]

Just to finish up.
here is the result of running the suggested RPM

rpm -e $(rpm -q kernel kernel-smp kernel-hugemem | sed /$(uname -r | sed -e s/smp//)/d)
error: package package is not installed
error: package kernel-hugemem is not installed
error: package is is not installed
error: package not is not installed
error: package installed is not installed

I am guessing that if the files that are being removed have already been removed, perhaps, by last  the successful yum update.
There are only 2 version groups in the /boot directory.

Thanks again for your help.