Hi,
Did you really have a nodeid = 100 or did I understand wrong here. Wouldn't expect that
to work very well.
There is ongoing work on documenting the start phases to improve understanding of what's
going on in the restart procedures.
Rgrds Mikael
Adam D wrote:
> I restored backup id 1 like so;
> In /var/lib/mysql-cluster/BACKUP/BACKUP-1/
> c1s1# /usr/local/mysql/bin/ndb_restore -n 10 -b 1
> -m
> c1s1# /usr/local/mysql/bin/ndb_restore -n 10 -b 1
> -r
> c1s2# /usr/local/mysql/bin/ndb_restore -n 20 -b 1
> -r
> c1s3# /usr/local/mysql/bin/ndb_restore -n 30 -b 1
> -r
> c1s4# /usr/local/mysql/bin/ndb_restore -n 40 -b 1
> -r
>
> My Node IDs are ndb( 10, 20, 30, 40 ) mgt (100)
> to keep things clear.
>
> I did that one at a time becuase my cluster config
> only allowed one extra api connection. Ive since
> changed this. However after carrying out this on
> all datanodes The cluster log mentioned
>
> 2005-07-15 11:04:28 ALERT -- Node 10: Network
> partitioning - arbitration required
> 2005-07-15 11:04:28 INFO -- Node 10:
> President restarts arbitration thread
> 2005-07-15 11:04:28 ALERT -- Node 10:
> Arbitration won - positive reply from node 1
>
>
> Then all the datanodes disconnected. I have no
> explicit abrbitrators set in my config.
> Apon starting them backup without --initial they
> came up, however stayed on Phase 4 for ages, and
> nodes '30' and '40' were cranking in Operations
> for quite some time;
>
> 2005-07-15 11:14:22 INFO -- Node 40:
> Operations=1975
> 2005-07-15 11:14:22 INFO -- Node 30:
> Operations=1975
> 2005-07-15 11:14:22 INFO -- Node 10:
> Operations=0
> 2005-07-15 11:14:22 INFO -- Node 20:
> Operations=0
>
>
> So my questions are;
>
> 1. Is that normal restoration behaviour, and;
>
> 2. Does anyone have a list of the Phases and what
> they do/mean so I can be clearer in the clusters
> operation.
Mikael Ronstrom
Senior Software Architect, MySQL AB
My blog:
http://mikaelronstrom.blogspot.com