MySQL Forums
Forum List  »  NDB clusters

Re: Got temporary error 899 'Rowid already allocated' from NDBCLUSTER
Posted by: Mikael Ronström
Date: May 18, 2017 05:27PM

899 is a temporary error that can happen in very rare cases.
It happens when the primary replica assigns a row id to the new
record and then when arriving at the backup replica the row id is not

This can happen in rare cases with special mixes of operations in the cluster.
The condition usually only stays for a very short time since it comes from
cases where the row id is released in the primary before it is released in
the backup replica.

We have worked hard to make this error extinct, but
there are still some very rare cases that pop up every now and then.
Every time we find such case we try to work out a solution for that
situation. So please report the problem as a bug and provide some
background to how it happened.

There is no special action needed to troubleshoot this.

No special action is needed, simply retry the transaction and it normally
works again. So it should not be necessary to restart any node. If for
some reason this problem continues and not even a node restart fixes it,
then an initial node restart fixes it since this copies data from the
surviving node and thus the replicas are synched up again.

I have however not any knowledge of the 899 error being that persistent
at any time. But temporary error 899 we have seen at times.

Options: ReplyQuote

Written By
Re: Got temporary error 899 'Rowid already allocated' from NDBCLUSTER
May 18, 2017 05:27PM

Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.