December 31, 20241 yr Previously, I had consistent transfers between disks in the array. They were about 200-250mb/s. However, after overcoming power issues, adding ZFS caches, and updating to 6.12.14, I've noticed that transferring files would start out fine, but then drop down to roughly 10-15mb/s, then pop back up after a short period of time. I don't recall this happening before, is there a cause someone could point me to? The diagnostics attached were taken during a file copy. To reproduce, I just copy a Windows ISO from an array share to another share on the array. Both are array only and do not touch the cache. All drives and SSDs (array and cache) are attached to a LSI 9300-16i SAS card. sarva-diagnostics-20241231-1250.zip Edited January 3, 20251 yr by S0ulDrag0n
January 1, 20251 yr Author I am not so sure. I just tried copying a file from my zfs cache pool to another folder in the same pool. I saw the same behavior where it would start high 200-300mb/s then drop to 5-15mb/s. I did notice somthing this time since it got stuck under 500kb/s. One core stayed at 100% the whole time: I was able to reproduce this a few times. I've attached a few screenshots I tried to capture while this was happening from top, htop, and iotop.
January 1, 20251 yr Author I've turned off my VMs and docker containers this time to see if I could capture a clearer picture. I was still able to reproduce the issue with everything off.
January 2, 20251 yr Author Thanks for your response Jorge. Please find the diagnostics taken during a large file transfer attached. The file transfer was of a 7.4GB file on a cache drive to another location on the same drive. It was also initiated through a share. I took two diagnostics: one at the start and one once it slowed down. I also took a third one since the second diagnostic was too close to the end of the file transfer while waiting for the first to finish. Something that also happened was that after the first file transfer, Unraid also locked up for several minutes and took awhile for the Web UI to respond. That was between the second and third diagnostics. This was never observed until now when I tried to get the diagnostics again. sarva-diagnostics-20250102-0947.zip sarva-diagnostics-20250102-0940.zip sarva-diagnostics-20250102-0939.zip Edited January 2, 20251 yr by S0ulDrag0n clarification
January 2, 20251 yr Community Expert I missed that you were copying into the same pool, copy from one pool to another, or from pool to array, or array to pool, if copying to the array enable turbo write first.
January 3, 20251 yr Author Here you go. Two diagnostics of when copying a large file from the pool to array. One right after start and one in the middle when it slows down. sarva-diagnostics-20250102-1901.zip sarva-diagnostics-20250102-1902.zip Edited January 3, 20251 yr by S0ulDrag0n
January 3, 20251 yr Community Expert Solution Diags show writes at around 80MB/s, which is pretty good without turbo write, try with turbo write enabled.
January 3, 20251 yr Author 10 hours ago, JorgeB said: Diags show writes at around 80MB/s, which is pretty good without turbo write, try with turbo write enabled. Hey Jorge, I get consistent speeds with turbo write enabled. This setting must have been enabled prior, but somehow got reset. The only recent event I could think of was an unclean shutdown during boot. I noticed the server name was reset, but nothing else had changed. If nothing else stands out to you, then I think I've finally gotten things in order. Thanks for your help! Edited January 3, 20251 yr by S0ulDrag0n
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.