Tracker not updating when mindevcount changed?

[Andy Lo A Foe]

The trackers cache the class mindevcount (and probably some other
params too) for a some time so unless you restart your tracker(s)
after config changes you might still see some old values.

To retriggger replication after changes:

$ mogadm settings set enable_rebalance 1

That, or do a mogadm fsck on your cluster..

Gr,
Andy

On Dec 14, 2007 5:54 PM, Clinton Goudie-Nice <cgoudie at twelvehorses.com> wrote:
>
>  Greetings all,
>
>  I'm having a weird problem in my testing, and I'm wondering if I'm doing
> something wrong.
>
>  I'm seeing this:
>
>
>  mogadm --tracker=prometheus:6001 class modify testdomain testclass
> --mindevcount=3
>  mogadm --tracker=prometheus:6001 class list
>
>  domain               class                mindevcount
>  -------------------- -------------------- -------------
>  testdomain           default                   2
>  testdomain           testclass                 3
>
>
>  That all looks good...
>
>  mogadm --tracker=prometheus:6001 stats
>
>  Statistics for replication...
>    domain               class        devcount      files
>    -------------------- ----------- ---------- ----------
>    testdomain           testclass           2      34476
>    -------------------- ----------- ---------- ----------
>
>  That doesn't look right to me, so I'll check the class list again...
>
>
>  mogadm --tracker=prometheus:6001 class list
>  domain               class                mindevcount
>  -------------------- -------------------- -------------
>  testdomain           default                   2
>  testdomain           testclass                 2
>
>  Wait, it's back to two? I'll check it again just to be sure....
>
>  mogadm --tracker=prometheus:6001 class list
>  domain               class                mindevcount
>  -------------------- -------------------- -------------
>  testdomain           default                   2
>  testdomain           testclass                 3
>
>  Huh?
>
>  If I keep running it, I'll see it flip back and forth every few seconds.
>
>  In the DB, it always shows correctly though:
>  select * from class;
>  +------+---------+-----------+-------------+------------+
>  | dmid | classid | classname | mindevcount | replpolicy |
>  +------+---------+-----------+-------------+------------+
>  |    1 |       1 | testclass |           3 | NULL       |
>  +------+---------+-----------+-------------+------------+
>
>
>  Finally, once I do get the mindevcount changed, how do I initiate the
> cluster replicating copies of everything it only has 2 copies of?
>
>  Thanks!
>
>  Clint
>
>  PS: I have 1 tracker, 4 storage nodes, each with a device.
>
>
>
>
>
>
>
>
>  Clinton Goudie-Nice
>  Architect / Senior Software Engineer
>  cgoudie at twelvehorses.com
>
>  Phone: +1.801.571.2665 ext 3264
>  Mobile: +1.801.915.0629
>  Fax: +1.801.571.2669
>  LinkedIn: http://www.linkedin.com/in/cgoudie
>  Twelve Horses
>  13961 Minuteman Drive
>  Suite 125
>  Draper, UT 84020
>  www.twelvehorses.com
>
>
>
>