[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: (ITS#4659) Core dump after MOD

--On Friday, September 01, 2006 9:44 AM +0200 Pierangelo Masarati 
<ando@sys-net.it> wrote:

> quanah@stanford.edu wrote:
>> --On Friday, September 01, 2006 1:40 AM +0200 Pierangelo Masarati
>> <ando@sys-net.it> wrote:
>>> I note that HEAD & re24 radically differ in this area; that specific
>>> assertion does no longer exist, and the signaling to the writewaiters is
>>> much more lightweight.  Could it be related to sucking in an incomplete
>>> or inconsistent set of diffs?
>> Always possible, but I don't think so...
>> That specific assertion looks like it exists to me.
> correct, the code differs in many points, but that logic didn't change.
> I'm trying to figure out if and how, in the close procedure for
> connections, there's any guarantee that a pending write is definitely
> tried and completed before actually closing, and who and if, in this
> case, resets the value of c_writewaiter.

After looking at the timing of everything, it is possible that I did kill 
-9 on a syncrepl consumer right in the middle of the MOD where the master 
died.  That may have triggered this core, if the timing was all right down 
to the nanosecond...


Quanah Gibson-Mount
Principal Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html