[Bucardo-general] Test Failures: Serialized Isolation
Greg Sabino Mullane
greg at endpoint.com
Wed Oct 24 03:29:26 UTC 2012
On Tue, Oct 23, 2012 at 05:33:57PM -0700, David E. Wheeler wrote:
> 1. Fix the bug that sends the samedb sync to the wrong KID; or
> 2. Fix the bug where the KID thinks it manages only one sync
Neither of those are accurate: a kid is only responsible for one
sync, full stop. However, because modern Postgres uses a listen/notify
with a payload system, we can reduce the number of listen/notifies
flying around and have a single "kid" listener. Hence, a kid will
receive notices aimed at any kid, but we then filter out only ones
that are addressed to us via the syncname. If the kid is doing anything
at all with another syncs messages, that's a serious bug.
...
> So maybe the KID responsible for the pgtest5 sync isn't running?
That would be my guess. I did not see $syncname output in that debug,
but it should be pretty explicitly stated a few times in the bucardo
logs, by both the controller and the kid, as to which sync the kid
is responsible for.
--
Greg Sabino Mullane greg at endpoint.com
End Point Corporation
PGP Key: 0x14964AC8
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 163 bytes
Desc: not available
URL: <https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20121023/1d44af1b/attachment.sig>
More information about the Bucardo-general
mailing list