maximum of nsrmmd processes?

[nickelback]

Hello all,

we have 8 tapedrives in use. In standby mode ther are 8 nsrmmd processes. When the drives are writing there is a second nsrmmd for each drive, in total 16. Sometimes we see more than 16. Is this a failure?

 

[mlmmilkyway]

hi,

I believe one is started for each backup session, I have 4 drives, two currently in use, 6 backup clients running and 6 nsrmmd processes.

 

[MostlyNuts]

Each save stream that gets started uses an nsrmmd process when writing to tape. When the stream is done, the accosiated nsrmmd goes away.

 

[605]

Sorry to disappoint you. There is exactly one nsrmmd process per enabled device. No matter whether it is actually active or not. No matter which device type it is - it applies to both tape and disk devices.

I have no other explanation why there are more nsrmmd processes running on the same node except it is the NW server itself which also shared the devices with itself (yes, you can configure this).


Each save stream will be monitored by one nsrexec (not nsrexecd) at the NW server. So this guy only lives during the backup.

 

[nickelback]

Thanks so far.
Yes, our drives are direct connected on server no other storage node. This explains why we should see in max 16 nsrmmd processes. But where does the other (17 or 18...) processes come from?
Any idea?

 

[605]

No, except for the fact that you share some drives with another storage node, i don't have any idea.

 

[denisfr]

Hello,

We've migrated to NW 7.1.3 this saturday (from 6.1.3), and we have this muliples processes even we have only 8 tapes (connected in FC between the server and the silo).

07/26/05 13:45:07 nsrmmd #32: Start nsrmmd #32, with PID 2433068, at HOST
07/26/05 15:00:06 nsrmmd #32: Start nsrmmd #32, with PID 647368, at HOST
07/26/05 20:00:33 nsrmmd #32: Start nsrmmd #32, with PID 1716398, at HOST
07/26/05 21:30:07 nsrmmd #32: Start nsrmmd #32, with PID 2637826, at HOST
07/26/05 21:30:09 nsrmmd #33: Start nsrmmd #33, with PID 1781916, at HOST
07/26/05 22:00:09 nsrmmd #32: Start nsrmmd #32, with PID 1192028, at HOST
07/26/05 22:00:11 nsrmmd #33: Start nsrmmd #33, with PID 1220650, at HOST
07/26/05 22:00:13 nsrmmd #34: Start nsrmmd #34, with PID 1917154, at HOST
07/26/05 22:00:17 nsrmmd #35: Start nsrmmd #35, with PID 1523784, at HOST
07/26/05 22:00:19 nsrmmd #36: Start nsrmmd #36, with PID 774206, at HOST
07/26/05 22:00:21 nsrmmd #37: Start nsrmmd #37, with PID 1409172, at HOST
07/26/05 22:03:50 nsrmmd #38: Start nsrmmd #38, with PID 1781760, at HOST
07/26/05 22:31:01 nsrmmd #38: Start nsrmmd #38, with PID 663654, at HOST
07/26/05 22:31:39 nsrmmd #39: Start nsrmmd #39, with PID 729170, at HOST
07/26/05 23:03:30 nsrmmd #39: Start nsrmmd #39, with PID 409702, at HOST
07/26/05 23:33:49 nsrmmd #40: Start nsrmmd #40, with PID 2338882, at HOST
07/26/05 23:34:01 nsrmmd #41: Start nsrmmd #41, with PID 2584660, at HOST
07/27/05 00:05:17 nsrmmd #41: Start nsrmmd #41, with PID 1650768, at HOST
07/27/05 00:05:40 nsrmmd #42: Start nsrmmd #42, with PID 479292, at HOST
07/27/05 00:05:49 nsrmmd #43: Start nsrmmd #43, with PID 2150538, at HOST
07/27/05 00:30:08 nsrmmd #43: Start nsrmmd #43, with PID 1192140, at HOST
07/27/05 00:30:10 nsrmmd #44: Start nsrmmd #44, with PID 946180, at HOST
07/27/05 00:30:16 nsrmmd #45: Start nsrmmd #45, with PID 1781824, at HOST
07/27/05 00:30:20 nsrmmd #46: Start nsrmmd #46, with PID 364722, at HOST
07/27/05 00:36:05 nsrmmd #46: Start nsrmmd #46, with PID 1916932, at HOST
07/27/05 00:37:41 nsrmmd #46: Start nsrmmd #46, with PID 1601544, at HOST
07/27/05 01:00:10 nsrmmd #47: Start nsrmmd #47, with PID 2293872, at HOST
07/27/05 01:02:47 nsrmmd #48: Start nsrmmd #48, with PID 1523788, at HOST
07/27/05 01:02:54 nsrmmd #49: Start nsrmmd #49, with PID 893164, at HOST
07/27/05 01:08:28 nsrmmd #49: Start nsrmmd #49, with PID 2576420, at HOST

After calling the support, they tell us it was a new thing in 7.1.x

Some explanation (in the legato 25570, but I don't kown where to find it):

A few general rules for assigning sessions to nsrmmd

Symptoms:

- Save set sessions do not appear to be assigned as expected to nsrmmd
- Additional helper nsrmmd processes are present
- Target per Session values appear to be limiting factor instead of a load balance factor
- File type device configured for other pool than current backup
- Any other backup device is configured that does not have a tape available for the current pool required

1) Save sessions can join an existing mmd (that already has some sessions assigned to it) or they can be assigned to a different one.

2) Once sessions are bound to an mmd, that mmd needs to complete the session.

3) Once an mmd is then bound to a device, it needs to complete before another mmd can use that device.

4) sessions are assigned to an mmd, up to the 'target sessions', before considering another mmd (ignoring pool rqmts).

5) if different values of 'target sessions' exist for devices on the same host, the lowest value is used for session-mmd assignment, until the mmd is bound to a device