sleepless2
Cadet
- Joined
- Jan 8, 2021
- Messages
- 1
Hello,
I would like to understand why my performance of a pool drops from usable to unusable as soon as larger data is written there.
I define this a little more precisely:
I currently have a Z2 pool of 5.2TB on 5 cheap 2.5 inch notebook disks with 2TB each.
I pass this pool to an esxi via iscsi.
At the beginning I had an acceptable performance. (A little worse than I had before with the raid controller - but the initial speed was acceptable since Truenas offers so much good for it).
Now I copied a larger file from an external PC on a VM. (SMB share on an Ubuntu: which runs on the esxi and uses the TrueNAS iscsi storage).
Since then the performance is unusable.
I simply performed a scrub in desperation. (which is normally performed by default anyway weekly).
After that the storage ran almost with the old speed again.
Now I simply copied a larger file (about 60GB) directly in the Ubuntu VM and bang, I was back to the unusable performance.
Unusable means: According to rsync it only writes with 500KB-1Mb/s.
During this a restart of other VM's is hardly feasible. It takes about 10 minutes to boot a simple Ubuntu - which was otherwise done in seconds.
Smart-tests do not give any errors - not even the detailed ones from the console. All hard disks are ok - otherwise the performance wouldn't be almost at the old one after a scrub.
Does anyone have any ideas? Are there any special esxi iscsi settings missing?
Here is a small setup overview (space-wise the pool is also still far from 50% utilized: A performance loss due to this is also excluded):
Would be nice to gather ideas - ideas for figuring out the bottleneck and where/who is creating that bottleneck.
Thank you :)
I would like to understand why my performance of a pool drops from usable to unusable as soon as larger data is written there.
I define this a little more precisely:
I currently have a Z2 pool of 5.2TB on 5 cheap 2.5 inch notebook disks with 2TB each.
I pass this pool to an esxi via iscsi.
At the beginning I had an acceptable performance. (A little worse than I had before with the raid controller - but the initial speed was acceptable since Truenas offers so much good for it).
Now I copied a larger file from an external PC on a VM. (SMB share on an Ubuntu: which runs on the esxi and uses the TrueNAS iscsi storage).
Since then the performance is unusable.
I simply performed a scrub in desperation. (which is normally performed by default anyway weekly).
After that the storage ran almost with the old speed again.
Now I simply copied a larger file (about 60GB) directly in the Ubuntu VM and bang, I was back to the unusable performance.
Unusable means: According to rsync it only writes with 500KB-1Mb/s.
During this a restart of other VM's is hardly feasible. It takes about 10 minutes to boot a simple Ubuntu - which was otherwise done in seconds.
Smart-tests do not give any errors - not even the detailed ones from the console. All hard disks are ok - otherwise the performance wouldn't be almost at the old one after a scrub.
Does anyone have any ideas? Are there any special esxi iscsi settings missing?
Here is a small setup overview (space-wise the pool is also still far from 50% utilized: A performance loss due to this is also excluded):
Would be nice to gather ideas - ideas for figuring out the bottleneck and where/who is creating that bottleneck.
Thank you :)
