need some help/advice

[Allineedis]

Posted this under hardware, since I do not suspect that drivepool itself is the issue. Hope this is correct.

 

I have 2 identical servers running drivepool, but 1 server started to slowly behave badly. Issue is that when copying from a source to that server, I am getting more and more CRC errors and transfer speed varies from 100MB/sec to 5MB/sec (sometimes network connection is even dropped). I use Terracopy to transfer files.

 

Now I am kinda lost where to start/continue troubleshooting.

 

What I did till now:

- clean inside and cards

- checked Sata cables

- swapped LAN cable to server

 

I do not suspect the NIC since read speed seems OK. Also switch an router are OK, since the other server is on the same hardware and woks flawless.

 

I am using a Supermicro AOC-SAT2-MV8 in that server as opposed to a RocketRaid R2220 card in the other. Could that be the issue? Can I simply switch cards (have a spare Rocket Raid somewhere), without moving files from the attached drives. Or is this not possible? I mean simply pull out the one put in the other, restart server.

 

On 3 of the 18 drives I have some SMART errors, does not look all to serious, but could this be part of the issue?

 

Memory test? Have not done that yet. What will this accomplish? Just now swapped the memory against a spare one I had lying around, and issue remains. So memory is not the issue.

 

Any other ideas?

 

Thanks for any help.

[Christopher (Drashna)]

I would recommend checking the event viewer on the server in question. Look for disk, ntfs, smb or other related errors in the "SYSTEM" event logs.

 

To be honest, I expect one or more of your disks are having issues.

And what errors are appearing in SMART? Some are more serious (and indicate more of an issue) than others.

 

As for swapping cards, it depends on how the controllers work. If they are just passing the disks through, then you may have no issues doing that.

[Allineedis]

OK, forgot to look in Event Viewer. Talk about knowing your basics.....

 

Did have a look and yes there is a consistent atapi error, Ide/IdePort7. Next to this I have one drive showing extensive SMART errors, large numbers of reallocation sectors, including pending. Hopefully this drive is located at port7 (always a pain, to find the associated hardware). Should be, I have numbered all drives according to port numbers an this is drive 8 ( so counting port0, port1, etc...).

 

Next to this I did send a dump of the Dashboard.exe file to Alex which hangs at 2GB memory usage. Do not know if they are related or two seperate issues.

 

Now I have to wait till Drivepool has removed the above mentioned drive which can take a while (2TB Hitachi).

 

Will report back if that helped/solved the issue.

[Allineedis]

Drive has been removed and as test I copied 200GB of data to the server. Everything works fine again, transfer speed around 95MB/sec, no CRC errors. Only the remote connection does not want to go up again, but that will be sorted. Did not know that a dying drive could cause a whole server to behave badly. I will pay closer attention to SMART errors.

 

That's why a combination of drivepool and scanner is a must!

[Christopher (Drashna)]

Glad to hear it!

 

And yeah, a bad drive can cause issues in the pool.