"John P . Looney" wrote:
>
> On Tue, Dec 19, 2000 at 03:18:45PM +0100, Emiliano mentioned:
> > > Smeg. Just got a segmentation fault message in the logs. I'll run it with
> > > httpd -X and see can I repeat it, and get a core dump...
> > What Midgard version are you on? I mean, when did you download it? Some
> > packages have seen subreleases since b7. Check out the download page.
>
> I downloaded the 1.4b7 SRPMs last week...I'm thinking that it could be
> a good idea to do a complete rebuild from CVS.
It probably is. The source packages have been updated, but the SRPMs
not,
and we've fixed two things last week that were the cause of erratic
segfaults.
> > > I've been trying to run httpd -X and it's not dumping core, even though
> > > ulimit -c is set to 100MB. And, I can't run it in gdb, because it gets
> > > sigpipes all over the place. Is there a cunning way of debugging apache ?
> > "Smeg" and "cunning". British?
>
> Irish. But a long-time red dwarf/blackadder fan, it seems :)
*grin* I was thinking the same. The series are no longer shown here,
which
is a crying shame. Appearantly there's a larger market for 'the bold
and the beautiful' and the 10.000 knock-offs thereof. And Big Brother.
As
a dutchmen, I feel I owe everyone an excuse for guilt-by-association
for that one.
> I'm wondering have I......ack.
>
> I was playing with a CVS version of repilgard to load in a recent version
> of the blobs (to fix the missing blobs thing). Of course, to get it
> working, I'd to set LD_PRELOAD to be the directory with the CVS version of
> the midgard libs...no wonder mod_midgard was crashing; it was running with
> the CVS libmidgard. When I restarted apache, it pick those up...
Aha. It would have slowly picked up on it anyway, since apache stops and
starts its kids regularly. But memory structures and APIs have not
changed
since b7, so I wonder why this would be a problem?
> Sigh. I'm still getting problems assigning groups to users, but that's
> likely because mod_midgard crashed while it was updating the tables, and
> things went funny. Nothing is ever easy.
I'm involved with another project concerning databases, and these people
are pretty vocal about DBMSsen that use separate servers, and in
particular
DBMSsen that lack transactions and foreign keys. MySQL in other words.
Does anyone know how the transaction model that was recently added to
MySQL
works? Could we just go about it extremely coursely and start a
transaction
before the page executes, and commit at wrapup? What would be the speed/
serialization implications?
Emile
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
