smart hanging on librpm/__memp_fget_rpmdb

[Axel Thimm]

On Wed, Nov 22, 2006 at 07:20:02AM -0500, Jeff Johnson wrote:
> 
> On Nov 22, 2006, at 6:55 AM, Axel Thimm wrote:
> 
> >On Wed, Nov 22, 2006 at 10:10:34AM +0000, John Pilkington wrote:
> >>
> >>>Mauricio Teixeira (netmask) <mteixeira at webset.net> added the  
> >>>comment:
> >>>
> >>>Is this still an issue? What version of smart/rpm?
> >>>
> >>I have not reported this to Axel, but smart (from atrpms) does
> >>occasionally hang for me during the final stage of the 'updating  
> >>cache'
> >>phase.  It works again after a reboot.  This is not a really serious
> >>problem for me, because I reboot frequently, but it would be  
> >>difficult
> >>in a production environment.
> >>
> >>version is smart-0.42-39.fc5.at.i386
> >>
> >>This has been happening for several weeks - perhaps since the 2.6.18
> >>kernel came in.  I don't think I can be more specific about when it
> >>first happened.
> >
> >Lately there seems to be a general regression on FC wrt rpm based
> >software, which affects apt, smart and yum all the like and
> >higher lever depsolvers based on these, too, of course.
> >
> >Although noone seems to be able to put a finger on any part of
> >rpm/beecrypt/db4 that didn't have any relevant changes something seems
> >to be triggering race situations. Some blamed yum-updatesd, others the
> >way (the new) yum opens/closes the rpmdb, but it seems to even happen
> >w/o any yum present and on both FC5 and FC6, so it's probably a
> >combination of kernel 2.6.18 latency/lock changes and rpm that
> >triggers this. But ATM this is just my best guess based on the
> >observations of rpm/apt/smart/yum bugs reported.
> 
> Whatever, make up your own FUD.

Jeff, please *read* more carefully, randomly insulting and jumping the
gun is not helping. I wrote that there are people reporting these bugs
on yum-free systems and on older FC systems that don't have the new
yum "improvements" ...

>      https://lists.dulug.duke.edu/pipermail/rpm-devel/2006-November/ 
> 001858.html
> 
> yum deciding to open an rpmdb thousands of times per run in order to
> catch signals is the problem. Wanna bet? The fix is well underway.
> 
> There's a further problem that snatching join keys out of a rpmdb,
> keeping them around for a while, and then attempting to retrieve
> headers using previously known good keys is going to cause further
> heartburn down the road a bit.
-- 
Axel.Thimm at ATrpms.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.labix.org/pipermail/smart-labix.org/attachments/20061122/c3e7ea56/attachment.pgp>