List:Cluster« Previous MessageNext Message »
From:Magnus Svensson Date:August 10 2004 7:56am
Subject:Re: cluster recovery scenario
View as plain text  
On Mon, 2004-08-09 at 22:04, Crouch, Luke H. wrote:
> when upgrading our switch, we accidentally turned off the switch that the cluster was
> running on while the cluster was still up. 
> anyway, then we put all the machines on the new switch and restarted them and tried
> to start up the cluster, but kept getting errors. we eventually did ndbd -i on each...and
> that worked, but of course, we lost our data that way. is there a guide or a place in the
> manual that explains how to recover when the entire cluster goes down?

When all of the nodes in the cluster are stopped a "system restart"
needs to be performed. The normal way to do this is to make sure all DB
nodes have been stopped and then start all DB nodes in sequence, quite
quickly. The nodes will then connect to each other and perform the
system restart procedure, which involves reading the data and logs from
disk. The time to perform system restart will depend on the amount of
data you have in your database. Please inform us of any problems you
encounter during this stage.

You can also increase the loglevel for startup by adding
LogLevelStartup=15 to the DB section of your config.ini file.

Best regards

> thanks,
> -L
> Luke Crouch 
> 918-461-5326 
> luke.crouch@stripped 
Magnus Svensson, Software Engineer
Office: +46 709 164 491

cluster recovery scenarioLuke H. Crouch9 Aug
  • Re: cluster recovery scenarioMagnus Svensson10 Aug