[Spread-users] regular crashes in spread 4.0.0rc2

[John Lane Schultz]

If I recall correctly, the easiest way to trigger the bug is to cause
a new daemon membership to be installed.  You can do this by starting/killing a
daemon or causing/healing a network partition.

The crash is caused by illegal memory accesses where something is
interpreted as a pointer when it shouldn't be, so it jumps somewhere
random in memory.  As with most illegal access bugs it is not
completely deterministic and can sometimes trash your stack/heap,
which can ruin your core file.

The hope is that there shouldn't be crashes in the release version,
but there will always be bugs in software this complex.  If you do run
into bugs, then we are all here to help fix the bug.

Cheers!
John

---
John Lane Schultz
Spread Concepts LLC
Phn: 443 838 2200 
Fax: 301 560 8875

Thursday, November 8, 2007, 2:51:43 PM, you wrote:

> On Thu, Nov 01, 2007 at 12:11:00PM -0400, John Lane Schultz wrote:
>> There was a bug in G_compare_proc_ids_by_conf in the release
>> candidates that was fixed before official release.  The best
>> solution would be to upgrade your installation to the official
>> 4.0.0 release as other problems were fixed too.

> Is there a scenario where I can reliably duplicate the buggy
> behavior in G_compare_proc_ids_by_conf() in the release candidate?

> I'm basically gettings two sets of core files: those that point to
> G_compare_proc_ids_by_conf() as the culprit, and others that just
> list memory addresses.  So I'd like to do some testing to suggest
> whether the same problem is causing all core dumps, or if there are
> other problems as well.

> In short, we're definitely going to upgrade, but we'd like to know
> if there will still be crashes or not.  I'd like to be able to
> determine this before the release, if possible.

> Thanks!
> Matt


> _______________________________________________
> Spread-users mailing list
> Spread-users at lists.spread.org
> http://lists.spread.org/mailman/listinfo/spread-users