RPM broken

[James Olin Oden]

On Sat, 3 Apr 2004, Martin Alderson wrote:

> After a reboot it seems to have fixed itself. Hopefully it was just a 
> one off bug of some sort and won't occur again.
>
A reboot fixes all lockups because locks are either done with futexes 
(basically a mutex of sorts which is held in memory), and the transaction 
lock is done via locking a file, which also is ultimately implemented in 
memory as far as I know.  

Wish you could have got a back trace, though, because I want to know if 
other people are still seeing the scriptlet "hang" as I am.  Thanks 
anyway.
 
> Also, is it common for up2date to just lock up? It seems to for me very 
> much so I just use the up2date applet thing and launch synaptic to 
> update. What are other people's views on this
>
Well, up2date can have bugs...don't know if any of the new code ever opens 
the rpmdb, and in some instances does not close it.  OTOH, updated, yum, 
aptget, and the rpm cli itself are all slaves to librpm so if anything 
goes wrong there you can be sure you will see them in those programs.

Cheers...james

P.S.  Since you have experience a hang, that have been the one I am 
looking at, if you had the time, there is a test harness you can run 
against rpm at rpm-devel.colug.net.   It runs against a copy of your 
system database with anoxious test packages, so its pretty safe to run
on your system.  If you would like to give this a whirl, you can email me
directly for any help on setting up the test harness and interpretting a 
hang if you see one.