Re: Error with "Read Backup"

[Mikael Ronström]

Jeff Steele Wrote:
-------------------------------------------------------
> Thanks again for your assistance.

No problem, it's fun to be working with real issues every now and then.
I mostly work on new features,
so it is nice to see some real world issues from time to time.

Also interesting to see what kind of issues user hits.

>
> I set all tables to READ_BACKUP=0, but the
> resource usage didn't return to normal. I then
> restarted the cluster. While the first data node
> was restarting, the second one crashed. I then
> restarted both of them and the cluster seemed
> fine. Resource usage stayed high until I ran
> "analyze tables". Then resource usage dropped, but
> I don't know if it was because of the analyze
> command or just a coincidence. I then made a few
> tables READ_BACKUP=1. At the moment, ndbmtd is
> only using 17% of CPU. That is compared to 500%
> earlier.
>

Yep, pretty sure that READ_BACKUP should only
help.

Sounds interesting, if you have any data that you can share
about what happened it could be interesting. Use the bug report
in that case.

Obviously ANALYZE TABLE can help if something was wrong
with statistics on tables.

I'll see if I can find out if the index statistics is affected
by ALTER TABLE REORG.

> Since CPU usage is low now, I'll leave
> NoOfFragmentLogParts alone for now.

Sounds like a good plan, it is not really an issue to keep
this forever if it works fine.