[debian-mysql] mysql replication issues since security upgrade

[Brent Clark]

To whom it may concern

I originally sent the below email.

Since then,we actually realised the problem was bigger than we anticipated. We thought it was a memory issue, so therefore I sent an email to the Mysql mailinglist ( Please see http://lists.mysql.com/mysql/227047 (see the stack trace too)), to try and deal with it.

But the problem still persisted.

We then thought it was faulty ram, we therefore replaced the hardware.

What's actually interesting is that we then setup another server with Mysql 5.1.49 (i.e. before the security upgrade).
We then got replication working, and then after a 'soak period', continued to upgrade to 5.1.61, and as a result we started getting 'mysqld got signal 6' messages.

I sincerely hope im making sense, for this had been carrying on for quite some time.

If you can help anyway, it would be appreciated.

Kindest Regards
Brent Clark

On 26/03/2012 15:51, Brent Clark wrote:
> To whom it may concern.
>
> We have been struggling with MySQL replication, why one node continuously crashes, replication has issues.
> On restore, when we make a snapshot of the master, we too get the following.
>
> Mar 26 14:43:54 xyz-web02 mysqld: 120326 14:43:54 [ERROR] /usr/sbin/mysqld: Table './db/comm_opens' is marked as crashed and should be repaired
> Mar 26 14:43:54 xyz-web02 mysqld: 120326 14:43:54 [Warning] Checking table: './db/comm_opens'
> Mar 26 14:43:54 xyz-web02 mysqld: 120326 14:43:54 [ERROR] 1 client is using or hasn't closed the table properly
>
> What we have identified, is that since about the 9 March, when we upgrade due to the following
>
> http://www.debian.org/security/2012/dsa-2429
>
> Do you think the two could be related?
>
> If you can help in anyway, it would be gratefully appreciated.
>
> Kindest Regards
> Brent Clark