MIB - Mandriva International Backports

Posted: **5 April 2013, 10:32**

I already reported this bug in the Rosa bugzilla at http://bugs.rosalinux.ru/show_bug.cgi?id=1882 and was told to try the MIB forum.

The problem is that MandrivaUpdate crashed with a SEGV while performing a regular update. Rosa Bugzilla seems to say that the problem may be caused by a big update when you run non-standard (MIB) repos, due to the fs running out of space. I had 700MB free, which seems plenty to me, and the following repos:

Code: Select all

# urpmq --list-media
Main_64_release
Main_64_updates
Contrib_64_release
Contrib_64_updates
NonFree_64_release
NonFree_64_updates
Restricted_64_release
Restricted_64_updates
MIB-basic_64
MIB-experts_64
Main_32_release
Main_32_updates
Contrib_32_release
Contrib_32_updates
NonFree_32_release
NonFree_32_updates
Restricted_32_release
Restricted_32_updates
MIB-basic_32
MIB-experts_32

It does not sound right to me. How can this morning's update of less than a dozen packages eat up 700MB after running for a minute or so?

Note that Drakbug also failed while reporting the above, but it's an unrelated problem due to Rosa Bugzilla not playing nice with it.

Posted: **5 April 2013, 11:20**

You can try:

- temporarely disable MIB /basic & /experts repo
- do official ROSA updates
- enable again MIB /basic & /experts repo
- do all other MIB updates

perhaps, spitting updates into two pieces, might be a workaround

you can try also using the konsole instead than gurpmi

Code: Select all

# urpmi --auto-select --auto

BTW, 700Mb of root / is a too small free space for / file system...
- you can try some app to clean all the unused and / or temporary files like bleachbit
- if you have few unused big app or game occupying a lot of root you could remove them before updating
- next time you install it's better to configure a bigger space for / > 12Gb (I ever configure at least 20Gb)

Posted: **5 April 2013, 12:01**

Thanks, the update resumed and finished after the crash. If I had suspected a low disk condition I would have selected fewer packages to update.

I accept that 700MB is too low for general use, especially for video consumption, but I still think it ought to suffice for a system update. I just can't imagine the system shoving 700MB around just for an rpm update, oh well. I've got lots of old servers with old mandrivas chugging along just fine with even fewer resources and I remember performing remote updates of up to 900 packages without problems! Now I have to worry about one more thing with these newer os'.

I always assumed the rpm subsystem already managed "big" (in this case only a half a dozen packets!) installs intelligently by breaking them up into smaller chunks - I remember installers doing that, so all the machinery is already in place. Then it would seem trivial to me to add this functionality (if it is not already there and the crash problem had another cause).

Posted: **3 June 2013, 13:06**

Update: it seems Rosa bugzilla fixed it, so I'll consider it solved. From their wording it does not seem related to "low" disk space (700MB for a few packages):

Comment # 2 from Denis Silakov

The problem should not rise anymore. We have also changed urpmi default
behavior to download all packages before installation, so users with default
configuration should not experience any problems if we perform similar big
updates in future.

MIB - Mandriva International Backports

[SOLVED] MandrivaUpdate SEGV's during regular update

[SOLVED] MandrivaUpdate SEGV's during regular update

Re: MandrivaUpdate SEGV's during regular update

Re: MandrivaUpdate SEGV's during regular update

Re: MandrivaUpdate SEGV's during regular update