Parity Re-Check speeds terrible ~3Mbps :(

[blade316]

Hey guys,

 

I'm hoping someone can offer some guidance here, I'm having a very strange problem.

My parity recheck speeds are atrocious most of the time, I get short spurts of high speeds then it goes back down again.

It stays around 1.5-3Mbps most of the time, so basically my UnRAID server is unusable during a parity check, when it jumps up it goes to 115-130Mbps, that lasts around 15mins maybe whenever it does it.

 

Major components

6 HDD's attached to SATA ports on motherboard

Two 8TB (one is the parity drive), Four 4TB drives

Core i7

24GB RAM

No cache drives

 

I have tried shutting down all my docker containers, but that made no difference at all .....

 

I've attached my diagnostic logs, I've had a look but cant really see anything that stands out, however I'm pretty green at reviewing these logs. I get the feeling its perhaps one or more of the drives causing the slowness, however I cant see anything in the SMART tests, and also as its just a check and not a rebuild, I believe its only reading the drives rather than writing to them?

 

unraid-diagnostics-20180318-2308.zip

 

Any assistance would be appreciated.

 

Cheers,

 

Dan

[Squid]

Does it make a difference if you shutdown your VM?  Does it make a difference if you shut down dynamix folder caching?

[JorgeB]

You have multiple reads/writes going on, stop all docker, vms, etc

[DZMM]

I haven't checked your logs, but not having any cache drives has got to hurt your parity check speeds as all new writes are going direct to the array, forcing the parity calc to change whilst it's trying to do a check

[blade316]

@Squid

 

6 hours ago, Squid said:

Does it make a difference if you shutdown your VM?  Does it make a difference if you shut down dynamix folder caching?

 

I shut down my VM and folder caching was already disabled, unfortunately it made no difference.

 

@johnnie.black

 

6 hours ago, johnnie.black said:

You have multiple reads/writes going on, stop all docker, vms, etc

 

When I woke up this morning it was going around 90mbps .... so still a lot slower than it should be, I disabled both the docker and VM services, but it didn't improve the speed .... I mean 90mbps is far better than 3mbps haha .... 

 

@DZMM

4 hours ago, DZMM said:

I haven't checked your logs, but not having any cache drives has got to hurt your parity check speeds as all new writes are going direct to the array, forcing the parity calc to change whilst it's trying to do a check

 

I think this is similar to johnnies reply, advising of read & writes while checking parity .... however stopping the docker and VM services didn't seem to improve the speed. ..... perhaps 90mbps is normal? And 130mbps only happens occasionally?

 

also it's just finished the check, and I've got 2 sync errors, never had those before. I'm wondering if it's got to to with the musicbrainz docker I tested over the weekend, as I had to create folders on a specific disk for it to work, rather than in /mnt/user

 

 

[JorgeB]

16 minutes ago, blade316 said:

When I woke up this morning it was going around 90mbps

If you mean 90MB/s, it might be normal at that point, disks slow down considerably as they go through the inner cylinders.

[blade316]

3 minutes ago, johnnie.black said:

If you mean 90MB/s, it might be normal at that point, disks slow down considerably as they go through the inner cylinders.

 

That would probably be why then, as it was at 97% of the check by the time I woke up this morning.