I am not aware of a way to force AutoBalance to a specific node pool, however....
Some guesses based on the info given:
1. Your job is being paused. I can think of 3 reasons why this might happen
a. FlexProtect is running (or ran for a good portion of that 60 hours). When FlexProtect runs, all other jobs are paused as data protection is given the highest priority.
b. There is another restriping job running at a higher priorty. By default this would only be MultiScan which has the same default priority as AutoBalance. If this is the case, you can tweak the priorities of the jobs to get AutoBalance to run.
c. You have 3 jobs already running with the same or higher priority than AutoBalance. Less likely, but technically possible. Again, you can tweak priorities to change this.
2. You have a ton of files to check. If you have SSDs in your cluster, you may try to run AutoBalanceLin instead of AutoBalance and it may go faster.. If you have older nodes without SSD, don't bother.
3. You've got a ton of load on your cluster. In this case, you could bump up the impact policy on the job (note this is different then priority), but it could take cycles away from users if you do so. Note: You can change this on the fly to run higher nights and weeknds, for example.
I ranked these in the order I think is most likely.
AdamFox
254 Posts
0
October 24th, 2017 07:00
I am not aware of a way to force AutoBalance to a specific node pool, however....
Some guesses based on the info given:
1. Your job is being paused. I can think of 3 reasons why this might happen
a. FlexProtect is running (or ran for a good portion of that 60 hours). When FlexProtect runs, all other jobs are paused as data protection is given the highest priority.
b. There is another restriping job running at a higher priorty. By default this would only be MultiScan which has the same default priority as AutoBalance. If this is the case, you can tweak the priorities of the jobs to get AutoBalance to run.
c. You have 3 jobs already running with the same or higher priority than AutoBalance. Less likely, but technically possible. Again, you can tweak priorities to change this.
2. You have a ton of files to check. If you have SSDs in your cluster, you may try to run AutoBalanceLin instead of AutoBalance and it may go faster.. If you have older nodes without SSD, don't bother.
3. You've got a ton of load on your cluster. In this case, you could bump up the impact policy on the job (note this is different then priority), but it could take cycles away from users if you do so. Note: You can change this on the fly to run higher nights and weeknds, for example.
I ranked these in the order I think is most likely.
Hope this is a start.