Upgrading now. Shutdown cluster. Upgraded all RPMs. Tried to come back
online. Crash. Some issue on node 1 with "file not found". Initial'd it
then brought nodes 2 and 3 up. They (thankfully!) started and are now
synching node 1. Waiting for that to finish before bringing node 4 up.
Noticed that in the individual ndbd node log files, I'd see lines like
[ndbd] INFO -- Start phase 5 completed
But when in the ndb_mgm, `all status` it always said "Last completed
phase 0" all the way until it 'Started'. That normal? I don't think so.
I always remember seeing the correct phase in the manager.
> -----Original Message-----
> From: Andrew Hutchings [mailto:andrew.hutchings@stripped]
> Sent: Friday, May 21, 2010 4:25 PM
> To: cluster@stripped
> Subject: Re: nodes ignoring MemReportFrequency setting when 80%
> Hello Matthew,
> On 21/05/10 21:58, Boehm, Matthew wrote:
> > I've just finished my test to confirm this. I loaded up some data
> > "all dump 1000" reported 79% data usage. In my ndb_mgmd config, I
> > specified MemReportFrequency=600 (every 10 minutes). As long as data
> > usage stayed below 80%, the usage was reported properly every 10
> > in the ndb_mgm console and thru the MGM API.
> > But as soon as that 1 record was inserted to push the usage to
> > the report frequency suddenly changed to every second for all 4 ndbd
> > nodes!
> > My mgm console was now being flooded with the reports as is the MGM
> > I don't see anything in the docs that state this is normal behavior.
> > possibly?
> There were a batch of fixes to memory reporting in 7.0.14 and 7.1.2.
> should have been fixed amongst those. 7.1.3 is GA now so if it still
> happens in this version please let us know.
> Kind Regards
> Andrew Hutchings, MySQL Support Engineer, Americas
> Oracle Corporation UK Limited.
> MySQL Cluster Mailing List
> For list archives: http://lists.mysql.com/cluster
> To unsubscribe: