FC6 - rpm --verifydb reveals problem

[Jim Cornette]

Tony Nelson wrote:
> At 7:04 AM -0500 11/20/06, Jim Cornette wrote:
> 
> Rebooting removes such lock files already.  The ones you removed are fresh
> ones, hopefully not still in use. 

Originally, I only removed the files before rebooting to attempt to 
follow-up with running rpm --rebuilddb. The process did not change for a 
very long time. I rebooted after the process never progressed for at 
least an hour.

> It is best to do such things in Single
> User mode (runlevel 1), so there isn't something else trying to use RPM. 

Agreed! I usually update the system after crashes in yum or rpm in 
runlevel 1.

> I suspect that the files are created by yum-updatesd or pup and that 
> they're "working on it".
> 
> I'm surprised that --rebuilddb worked.  Maybe it only worked up to the
> point in Packages that has the error.  Your RPM database may be incomplete.
> You could compare it to before by looking at /var/log/rpmpkgs* files.

I'm in development version now. I'll compare the output between rpm and 
the logs you reference above once back in FC6. I compared the two in 
development and both matched up to each other. Of coarse the crash was 
in FC6, not devel.

> 
> 
>> The problem seems to be that once an error is encountered in the
>> database, you cannot run any instance of yum or any rpm instance with
>> options until you remove the /var/lib/rpm/__db.* files and reboot the
>> system. I believe the rebooting of the system either clears memory where
>> there is problems or at least clears the system from temporary files
>> which are causing the lock.
>  ...
> 
> I have those files but I can still use yum and rpm.

I don't buy my statements above after reading through them either. The 
symptom that once a crash happens, nothing can be done with rpm or yum 
until a reboot is true though.

Thanks for the tips.

Jim

-- 
Support staff hung over, send aspirin and come back LATER.