[torquedev] pbs_server segfault in req_delete.c

Joshua Bernstein jbernstein at penguincomputing.com
Tue Dec 30 12:10:43 MST 2008



Josh Butikofer wrote:
>  > What do you think Josh (CRI Josh, not Penguin Josh!)?  I think we
>  > would want to upgrade the bugzilla at
>  > www.clusterresources.com/bugzilla and probably flush out all the old
>  > bugs and start fresh since TORQUE has changed so much since they were
>  > loggeed.
> 
> I think it could be a good idea. The only problem that I see is the fact 
> that there would then be potentially three sources for bug 
> tracking/reports:
> 
> * The mailing list (when someone doesn't add it to Bugzilla)
> * The public Bugzilla
> * CRI's own internal ticket system (where we track most bugs that we are 
> working on)
> 
> It could make tracking bugs even more complicated, rather than easier.

I totally agree, though I think there has to be some sort of guidelines 
here. Ideally I think there would only be a small few of us that could 
add bugs to Bugzilla. Though anybody would be allowed to search through 
it. That why when somebody posts to the mailing list, one of say 3 or 4 
of us would descide its an actual bug, and could file a report.

As for the CRI side, I would say that in the spirit of the community, 
the public bugs and the "private" bugs should be in the same place. 
TORQUE is a community project (right?) and thus I don't think anything 
should be hidden from the community. That said, I do understand that 
perhaps some business case somewhere won't allow real transparency. 
Perhaps there is a hook inside of Bugzilla that would allow the sinking 
of the public bugs to the internal system?

> As long as CRI and the community have clear rules about how to keep 
> Bugzilla up-to-date and accurate it should work fine. The problem with 
> the old Bugzilla (from my perspective) is that only a few dedicated 
> members of the community used it for a time, and as the number of bugs 
> built up, and TORQUE fixes were slow to come, people started realizing 
> they could get better turn around time and responses if they just 
> squawked on the mailing list, bypassing the whole Bugzilla tracking. 
> Eventually, it became moth-balled and we are where we are today, with no 
> one really using it.
> 
> My biggest fear, I guess, is that Bugzilla, like the TORQUE wiki, will 
> not be used enough to warrant the time and resources put into 
> maintaining it. Is TORQUE really isn't going to benefit from having 
> Bugzilla? How can we avoid the problems that occurred last time?

We just have to come up with some sort of agreed upon guidelines. I'm 
gotten permission internally for me to spend some real time on the 
TORQUE project, so as I mentioned previously I'm happy to volunteer for 
whatever needs to be done.

-Josh Be



More information about the torquedev mailing list