Bug #22245 Cluster crashes, won't restart
Submitted: 11 Sep 2006 19:57 Modified: 22 Sep 2006 2:24
Reporter: Jeff Tucker Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S1 (Critical)
Version:5.0.22 OS:Linux (Linux)
Assigned to: CPU Architecture:Any
Tags: cluster, ndb

[11 Sep 2006 19:57] Jeff Tucker
When running, one ndb node eventually crashes. The other node currently won't start so I'm trying to use only a single node.

On restart, this node again immediately crashes.

How to repeat:
[11 Sep 2006 20:08] Jonas Oreland
Please upload trace/error files and cluster log + config.ini
[11 Sep 2006 21:00] Jeff Tucker
I did upload the files. Are there other files you need to see?
[11 Sep 2006 21:23] Jonas Oreland

This looks like http://bugs.mysql.com/bug.php?id=21384
which has been fixed in 4.1.22/5.0.25/5.1.12

Have you tried starting the node "--initial"
This is perfectly safe, as all data is also in running node
  (if you only do this on _1_ node)

[11 Sep 2006 21:35] Jeff Tucker
Thanks, Jonas. Unfortunately, for whatever reason, now the other node is crashing, too. Basically, neither node will come up at this point and the cluster is down.

Luckily, during a moment when it was running I did a dump of all the data. I'm currently running a non-cluster MySQL node so I'm back up and running. 

At this point, I've got two nodes which have both crashed. I don't believe I can restart either by itself and it sounds like using --initial in this case would be a bad idea.
[12 Sep 2006 8:39] Jonas Oreland

As this really looks like bug that I refered to which is fixed in 5.0.24
I would recommend upgrading.
And restarting the new cluster with entirely clean filesystem
And then restore data into cluster using sql dump.

[22 Sep 2006 2:24] Jeff Tucker
Thanks for the help. I'll figure out where to go from here.