Eric,
SEQUOIA-894 seems to be a different issue since it was for pending
transactions not aborted, not pending writes like you reported. Also
this use case is not produced by a garbage query but by a thread crash
in MySQL on only one of the backends. This does not look like the
problem you reported initially.
As for 3.0, this branch has been obsoleted. I am the only one working on
Sequoia 4 for now. Continuent might release a 2.10.11 version for their
next product release but I have no idea if and when this is going to
happen. I don't have commit access anymore to 2.10 and cannot contribute
fixes to that branch anymore.
Don't hesitate to file a new JIRA entry with the specifics of your use case.
Emmanuel
After digging through JIRA I found there is already a bug that matches what I
am experiencing. https://forge.continuent.org/jira/browse/SEQUOIA-894
Is there a release date for 3.0 or 2.10.11?
-Eric
________________________________________
From: [EMAIL PROTECTED] [EMAIL PROTECTED] On Behalf Of Emmanuel Cecchet [EMAIL
PROTECTED]
Sent: Wednesday, November 12, 2008 4:27 PM
To: Sequoia general mailing list
Subject: Re: [Sequoia] Failover Failure?
Which version of Sequoia are you using?
What database backends are configured? What version? What drivers?
What query breaks? Any error or warning in full_cluster.log?
Is the query in a transaction or in autocommit?
How is query executed (Statement, PreparedStatement, CallableStatement,
execute, executeQuery, executeUpdate, batch...)?
If you have a use case to reproduce the issue, please file it in a new
JIRA entry.
Thanks for your feedback,
Emmanuel
Eric Faden wrote:
As it turns out my backends were getting hung on shutdown, so my DB was in a
"Suspending" state. Both controllers just keep printing out Waiting for 1
pending writes. Looking at the queues show a broken query stuck in there with no way to
get it out (abort transaction didn't work). I had to force the DB down and then recover
it. I seem to be able to replicate this by shoving a broken query into the backends,
causes everything to get stuck open. Is there a fix for this?
-Eric
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Emmanuel Cecchet
Sent: Wednesday, November 12, 2008 4:10 PM
To: Sequoia general mailing list
Subject: Re: [Sequoia] Failover Failure?
Hi Eric,
Can you check that the backend on the active controller is still enabled?
What JDBC URL do you use in your application?
Thanks for your interest in Sequoia,
Emmanuel
I currently have two machines, each with a MySQL Instance and
Sequoia instance. They are configured such that the controllers
mirror the two MySQL Instance (box 1 has box 1's mysql as backend,
etc). When I shutdown the virtual database on box 1 or box 2 the
entire cluster now fails to function or write any data. I get the error
"create table test (I int)/' failed on all controllers. If I bring
back up the second controller it all works again. Any ideas?
-Eric
------------------------------------------------------------------------
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
