#Imbalance in storage node capacity

1 messages · Page 1 of 1 (latest)

hushed forge
#

Hi all,
about 1 year ago, we did an expansion of our SG platform from 4 to 6 storage nodes. We're using it as a cloud tier for fabricpool. After adding the 2 additional nodes, we modified the erasure coding schema from 2+1 to 4+1. We noticed that capacity started to slowly rebalance across all nodes, and by the end of march 2023 all nodes were using more or less the same capacity.

During the same period, we added 1 more tenant and 2 more buckets, for different fabricpools.

Now looking at the capacity graphs we can see that in the middle of march 2023 used capacity has dropped for 4 nodes, meanwhile it raised for 2 nodes.

Should we worry about that imbalance?
What could be the cause of this behavior? I've checked already that ILM policy is correctly applied via simulate.

SG version 11.6.0.5

Thank you

frosty valley
#

there's not really a lot of imbalance here... make sure all the buckets/tenants are using the correct EC groups...

#

the curves are following each other pretty closely so it's probably not much to worry about... you probably should try to get to 11.7.04 (or .05 if it is ever released) sooner than later

ember void
#

Agree, this is not much unbalanced. I have seen it before that the nodes where newly added/emtpy at the time the EC conversion started end up slightly "more used" that the old ones. This nothing to worry about and can be explained how EC works. If the usage on the nodes will get close to 90% the other nodes that are slightly behind will catch up.

I also recommend to update the grid, but 11.6.0.13 (mandetory hotfix) would be enough. Likely the E-Series Santricity OS is also rather old on the Appliances, recommending to go to 11.70.5P1, a good release - at some point you need to go there anyhow.

summer path
#

@hushed forge or anybody here,

What steps do I need to follow to pull out the graph as you did "NetApp StorageGrid Node Space Usage"?

hushed forge
frosty valley