Array Slowdown

[LindsayCole]

Heyas

 

I have a 27.1 TB array, consolidated through DrivePool.

 

I suspect I have something lagging the array down. I am getting intermittent drops in performance (this is primarily a Plex server) and am having trouble nailing down the reason behind the slow down. I'm trying to nail down if I have a problem with the storage subsystem.

 

Is there an easy way to test each drive for speeds?

[saitoh183]

I would use StableBit Scanner to see if there is a issue with one of your disk. It can check smart status and few other stuff

[RobbieH]

Yep, a failing drive is exactly what caused my server to slow down. Found it with Scanner.

[Christopher (Drashna)]

Heyas

 

I have a 27.1 TB array, consolidated through DrivePool.

 

I suspect I have something lagging the array down. I am getting intermittent drops in performance (this is primarily a Plex server) and am having trouble nailing down the reason behind the slow down. I'm trying to nail down if I have a problem with the storage subsystem.

 

Is there an easy way to test each drive for speeds?

StableBit Scanner has the "Burst Test" option on each disk (right click on the disk in Scanner to see it).

This is very useful for identify stability or performance issues.

 

Additionally, the SMART data may indicate an issue. A high reallocated sector count will cause performance issues. As will a growing uncorrectable/pending sector count.

 

Baring that, the Event Viewer may reveal problematic disks. Check the "System" section for disk related issues. If you see a lot on a specific disk, then that disk may be having issues that are causing the performance problems.

 

 

 

Additionally, if this is happening when DrivePool is running, then could you enable file system logging and run it until the performance issues hit (and note the approximate time:

DrivePool 1.X: http://wiki.covecube.com/StableBit_DrivePool_Error_Reports

DrivePool 2.X: http://wiki.covecube.com/StableBit_DrivePool_2.x_Log_Collection

[LindsayCole]

Burst Test - I tested all drives. (for a 10 min period admittedly)- they all tested 210+.

 

A bit of insight into my setup;

 

Server 2012 R2

Dell C1100

24GB RAM

2x Xeon L5630s

IBM 1015 flashed to IR mode.

SE3016 SAS Enclosure

Plex Media Server

Sonarr

uTorrent

DrivePool / Scanner

 

I could have sworn I had access to the SMART data a month ago, checking now- I have nothing. I recently learned that the 1015 uses SAT (SCSI to ATA Translation) - does DrivePool support SAT?

 

 

Thanks!

[lee1978]

You may need to set scanner to another setting off the top of my head it is unsafe direct or something I think it's under the advanced settings sorry for being vague i don't have access to the server right now to check

[Christopher (Drashna)]

We absolutely support SAT for the IBM ServeRAID cards.

http://wiki.covecube.com/StableBit_Scanner_Advanced_Settings

Find the "DirectIO" section, and then check the "Unsafe" option.

You'll need to restart the "StableBit Scanner Service" (run "services.msc" to do so), or reboot the system.

 

 

 

As for the performance, it may take running for longer to really determine if/where the issue is.

However, have you checked the event viewer for disk related issues at about the same time that the performance issues crop up?

Also, is anything else running at the times that the performance tanks? Such as unpacking downloads, transcoding, etc?

 

And in DrivePool, is the "Network IO Boost" option enabled?