Hi Guys,

 

Diagnostics are attached. Rebuilding a drive, after upgrading to a larger drive, and I'm only seeing KB/sec for the first 45 minutes. Already rebooted, once, and restarted. First one ran 45 minutes at that speed. This time, I'm 25 minutes it at same slow speed. 

 

What am I missing?

tower-diagnostics-20180411-1037.zip

First thing would be to stop writes to other disks, there are simultaneous writes going on disks 1, 2, 3 and 18

2 hours ago, johnnie.black said:

First thing would be to stop writes to other disks, there are simultaneous writes going on disks 1, 2, 3 and 18

 

So I'm stumped. I don't have anything writing to the array. There's not much on my cache and mover doesn't appear to be moving. I've turned off my PCs just in case they were going rogue and writing, without my knowledge. 

 

Here is a new diagnostics, after a reboot. 

 

Seeing anything helpful?

tower-diagnostics-20180411-1329.zip

Still activity, looks like dockers, emby, radar, etc.

21 minutes ago, johnnie.black said:

Still activity, looks like dockers, emby, radar, etc.

 

I stopped all dockers and then stopped the docker utility, directly. Any better? no speed improvement

tower-diagnostics-20180411-1405.zip

Extra activity stopped, but if there's no speed improvement that wasn't the problem, there's no disk related errors on the logs, run diskspeed on all disks to see if speeds are normal.

1 hour ago, johnnie.black said:

Extra activity stopped, but if there's no speed improvement that wasn't the problem, there's no disk related errors on the logs, run diskspeed on all disks to see if speeds are normal.

 

OK, ran to completion the second time. Results are attached. Your help is greatly appreciated!! 

 

Edited April 11, 20188 yr by newoski

NOW it is running at 90MB/sec. WTF and also phew!

 

( :

1 hour ago, newoski said:

WTF

 

You have Archive drives in your array, including your two parity disks. With dockers writing to four disks simultaneously the parity updates are going to consist of random writes so the persistent cache is going to be involved. Perhaps it filled up and caused disk I/O to stall until it had emptied. Now it has emptied the speed has been restored.

 

That's my theory anyway, and my reason for arguing against using Archive drives for parity.

 
You have Archive drives in your array, including your two parity disks. With dockers writing to four disks simultaneously the parity updates are going to consist of random writes so the persistent cache is going to be involved. Perhaps it filled up and caused disk I/O to stall until it had emptied. Now it has emptied the speed has been restored.
 
That's my theory anyway, and my reason for arguing against using Archive drives for parity.

Interesting! Will consider swapping out w regular, go forward. Thanks so much for the knowledge

Sent from my HTC6545LVW using Tapatalk