List:Cluster« Previous MessageNext Message »
From:Jonas Oreland Date:October 12 2004 5:33am
Subject:Re: Major issues with NDB
View as plain text  
Hi Bruce.

Bruce Robertson wrote:
> Hi!  I'm running the daily snapshot from 09/25, and I've run into some major
> issues that I can't seem to recover from.  My config file is appended.
> 
> One of my programmers attempted to do an "alter table" to add a column.  That
> caused the cluster to crash, and now I can't get it running again.  Symptoms:

Is this bug reproducable?
Can you file it as a bug report at http://bugs.mysql.com ?

> I can start three of the DB nodes, though in spite of the config, the three
> all think they're in the same Nodegroup.  When I start the fourth, the entire
> cluster crashes again.  See log snippet below.
 >
> Any light you can shed would be great.  I'm happy to try a newer snapshot, but
> I'm afraid my ndb_X_fs directoies may already be corrupted beyond repair.
> I have the trace files, etc if you need them.

Here's an idea that will hopefully make you restart your cluster
without loosing the data.

Try performing the system restart wo/ node 3 (the node that fails first)

The can under most circumstances start wo/ all nodes preset.
Node 3 can in your case later be started with "ndb --initial", any join 
the cluster wo/ using it's own filesystem.

Question: What make you think that "think they're in the same Nodegroup"?
The printouts in ndb_mgm w.r.t nodegroup is not valid during startphase 
1 & 2 (actully it gets set during startphase 2).

/Jonas

-- 
Jonas Oreland, Software Engineer
MySQL AB, www.mysql.com
Thread
Major issues with NDBBruce Robertson12 Oct
  • Re: Major issues with NDBJonas Oreland12 Oct