03-14-2014 12:03 PM
i'm running 5220 appliance with 2.5.1, netbackup 7.5.4 on linux. we are backing to to disk.
we have a policy that quieries a VM cluster and pulls in many servers. They start running but after about the 10th one the snapshot details will state" Info nbrb (pid=???) Limit has been reached for the logical resource primaster.VMware.Datastore.vmwplz-XIV02-LUN14.
this happens on a lot of them but will eventually continue. I have worked with my support and changed number of jobs from 20 to 40 and on my netbackup master & media servers. I have changed the parms NUMBER_DATA_BUFFERS_DISK and NUMBER_DATA_BUFFERS_DISK. 64 and 1048576 , 64 and 262144, 128 and 1048576 , 128 and 262144 combinations.
i have moved policies to stagger between the master & media appliances to balance but some servers take the same amount of time no matter which appliance.
i cant see any changes in time or performance and wanted to know what this message may point to :Limit has been reached for the logical resource primaster.VMware.Datastore.vmwplz-XIV02-LUN14.
also i ran a test during the day when of course nothing else was running and the time cut in half. But the test only selected this server.
i'm wondering if instead of 1 policy selecting all VM's, if breaking them into 2 policies , or is something in VM .
i'm looking for guidance on where to find the bottleneck.
thanks
Solved! Go to Solution.
03-17-2014 08:01 AM
As long as your VMware policies are query based then there is no need to restrict the policy - the reource limits will do what you need
And yes, setting the datastore limit to 4 will allow 12 jobs at all times provided the policy / storage units etc can do that many jobs
So try 4 first, then try a lower value and see if the total backup window for the VMware jobs actually reduces
If you do not use a VMware policy type and a query for selecting the clients then the resource limits will have no effect
Hope this helps
03-16-2014 10:17 AM
did you tried limiting vmware jobs per datastore from master server host properties?
http://www.symantec.com/docs/HOWTO44509
Check there, also have look at your datastore I/O on ESXi
Also increase maximum number of jobs per client to 10-15 or even more as per your environment. You can set this from netbackup master server global attributes
03-17-2014 03:46 AM
This should be a Master Server Host properties setting as Captain Jack says above ... but having said that the reccomended optimum setting is just one per datastore - so if you are hitting 10 per datastore your system is probably not running as fast as it can do!
Better to limit it to a lower value .. they may queue up but the overall time to do all of the backups can actually be faster doing them one or two at a time that trying to do 10 or more at a time.
Hope this helps
03-17-2014 07:16 AM
I have looked at the "resource limit" tab , not really understanding what was best, and the link above was very helpful. But i have more questions because i dont know too much about VM. From what i understand about our VM, we have a 3 datastores, all in 1 cluster, that have many sevrers in each datastore. Under my resource limit, I have for "Datastore" a limit of 4. Everything else is "no limit".
would this be that i am allowing up to 12 Simultaniously backups. 4 from each datastore?
the other link stated best results, is that no more than 4 per datastore, which is why i am guessing it is set at 4. then it stated to "limit jobs per policy" would help.
Where i am getting confused is the "resource limits" tab looks like you can control limits at different levels. I dont limit "jobs per policy" now, but If I have 4 per datastore then the "jobs per policy" should not allow more than 4, Correct?
and would a lower number per datastore like 2, make it run faster?
thanks
03-17-2014 08:01 AM
As long as your VMware policies are query based then there is no need to restrict the policy - the reource limits will do what you need
And yes, setting the datastore limit to 4 will allow 12 jobs at all times provided the policy / storage units etc can do that many jobs
So try 4 first, then try a lower value and see if the total backup window for the VMware jobs actually reduces
If you do not use a VMware policy type and a query for selecting the clients then the resource limits will have no effect
Hope this helps