My parity checks have always clocked in at 129 MB/sec.  It's always been surprisingly consistent, but this weeks parity check has me a bit worried.  It took a bit longer and clocked in at 108MB/sec.

 

Is this indicative of an immanent drive failure?  The only thing different I've done recently is moved Plex Media Server onto the unRAID server (it was on a standalone server before) and I added DDclient, but those are located on my cache drive so I doubt that is causing this.  I don't think I was accessing the drives at the time, but even if I was, I've looked back and the speed has never not been 129MB/sec, even when I know I was accessing the drives.

 

Any ideas?

3 hours ago, Herdo said:

My parity checks have always clocked in at 129 MB/sec.  It's always been surprisingly consistent, but this weeks parity check has me a bit worried.  It took a bit longer and clocked in at 108MB/sec.

 

Is this indicative of an immanent drive failure?  The only thing different I've done recently is moved Plex Media Server onto the unRAID server (it was on a standalone server before) and I added DDclient, but those are located on my cache drive so I doubt that is causing this.  I don't think I was accessing the drives at the time, but even if I was, I've looked back and the speed has never not been 129MB/sec, even when I know I was accessing the drives.

 

Any ideas?

Seems like a pretty reasonable speed to me but I don't run any 7200 drives. We can't tell you anything else without more information of course. Tools - Diagnostics, post complete zip.

You can try to run this

and check if a disk is performing below par.

Thanks for the replies you two.  

 

trurl, I've included the diagnostics I took last night.  I looked it over and didn't see anything interesting.  Don't mind the morons trying to brute force my server.  I had port 22 opened for a few hours to test something and they came pouring in, haha.

 

dikkiedirk, thanks for linking that script.  wget and curl failed me (I assume due to the forum requiring login credentials for downloads) so I had to use scp.  All the disks seem to be performing just fine at about 137MB/sec on average.  I've included the diskspeed.html graph file if you want to take a look at it. 

 

I guess it's possible I was doing something weird during the parity check this week.  I think I might shut everything off that could potentially access the disk tonight and run another parity check.  So far it looks like this might just be some one off situation.

 

Thanks again

unraid-diagnostics-20170323-0128.zip

diskspeed.html

Everything looks good, tunables are not optimal but that wouldn't explain the slowdown, unless maybe it coincided with a new unRAID release, still you can try changing to these and run another check to see if it improves.

 

Settings -> Disk Settings:

 

md_num_stripes: 4096
md_sync_window: 2048
md_sync_thresh: 2000

25 minutes ago, Herdo said:

Don't mind the morons trying to brute force my server.  I had port 22 opened for a few hours to test something and they came pouring in, haha.

That will always happen, they are bots. No actual morons required.

28 minutes ago, johnnie.black said:

Everything looks good, tunables are not optimal but that wouldn't explain the slowdown, unless maybe it coincided with a new unRAID release, still you can try changing to these and run another check to see if it improves.

 

Settings -> Disk Settings:

 

md_num_stripes: 4096
md_sync_window: 2048
md_sync_thresh: 2000

Hey thanks.  The settings you gave me didn't make any difference with the diskspeed script, but maybe it will with the actual parity check.  I'll play around with it some more.

 

trurl, yea, this is why I use key pairs.

6 minutes ago, Herdo said:

The settings you gave me didn't make any difference with the diskspeed script

 

They wont make any difference there, that's your disks max speed, it may make some during a parity check.

Just now, johnnie.black said:

 

They wont make any difference there, that's your disks max speed, it may make some during a parity check.

 

Oh that makes sense, thank you.  I'll run another parity check tonight.