Highlighted
8 Krypton

Rebalance of large pools

Jump to solution

We just added more capacity and previously rebalances were pretty quick.  With this upgrade we added around 300 drives to our FC disk group.  Our FC disk group now has 1116 disks and 4464 devices.  The Max devs pre rebalance scan is set to the max (1024).  After approx 24hrs I get a unisphere alert that the rebalance completed but due to timeout.  When I check at the command line the pool state still says Balancing but when I run down the list of devices there aren't any that have a state of Balancing.  Some of the new devices have definitely been used but there are many that have not been allocated at all.  I tried a start and stop a few times over the past week but I still can't get this pool to even out.  Additionally devs that were at a high Full % were allocated additional tracks instead of utilizing the new empty devs.

We are at 5876.251.161 and utilizing FASTVP.  I also noticed that FASTVP seems to want to promote more data to the FC tier after this code upgrade and i'm not sure if that has anything to do with this rebalance.  Should I stop FASTVP movements to see if that helps to get this FC pool rebalanced?

Thanks

Labels (1)
1 Solution

Accepted Solutions
8 Krypton

Re: Rebalance of large pools

Jump to solution

FYI:  This is being resolved through a SR, but for everyone else this is a known issue with a change introduced in 5876.229.145 and 5875.267.221

KB 165917 and KB 172072

The work around is to run ECBB scans on the affected devices and after they are scanned they will be available for new allocations.

0 Kudos
4 Replies
8 Krypton

Re: Rebalance of large pools

Jump to solution

I had noticed the same behavior recently when I was trying to balance our pool. The pool state would show balancing with no data devs in the balancing state.

It took 4 days to rebalance the 120 TB SATA pool.

regards,

Saurabh

0 Kudos
8 Krypton

Re: Rebalance of large pools

Jump to solution

did you do anything to get the balancing going when you saw the devs weren't actually balancing?

0 Kudos
8 Krypton

Re: Rebalance of large pools

Jump to solution

No, i did not. It appeared that the rebalance runs randomly on the datadevs or atleast the output showed that..

0 Kudos
8 Krypton

Re: Rebalance of large pools

Jump to solution

FYI:  This is being resolved through a SR, but for everyone else this is a known issue with a change introduced in 5876.229.145 and 5875.267.221

KB 165917 and KB 172072

The work around is to run ECBB scans on the affected devices and after they are scanned they will be available for new allocations.

0 Kudos