Rebuild taking an inordinately long time

[Phil C.]

I was away on a trip when I had to talk my wife through shutting down my unRAID server due to a high temp alarm.  Upon my return (Saturday), I had to reboot several times to get the server on the network, then it showed Disk 1 as Disabled.  So I shut down, checked cables, rebooted, and started the Disk 1 rebuild.  It is now Wednesday, and the rebuild is only at 39% at a rate of 1,873 kb/sec with over 10,000 minutes left.  Is there anything I can do?  I'm willing to rebuild the parity vice the disk - if there is an error on Disk 1 it's no big deal.  Anything to get the array back up faster?

[prostuff1]

See the sticky at the top of this forum with instructions to include your unRAID version and syslog.  Without those we have not a clue on were to really start looking.

[Rajahal]

Was the high temp alarm from a fan controller, or was it related to the HDD temp?

 

Are you sure that the disabled disk was due to a loose cable and not actual disk failure?  Interrupting the data rebuild puts your data at risk.

 

You can cancel the data rebuild if you like.  unRAID won't throw away the disk 1 data until you tell it to.  What I would probably do in this situation is to just buy a new disk to replace disk1, let unRAID rebuild on to it and get the array back up (and parity protected) ASAP.  Then I would take my time evaluating your old disk1 to make sure it is actually trustworthy.  If it is, then you can add it as a new data disk.  If not, RMA it.

[Phil C.]

Running 4.7

Syslog attached.

 

More details - probably extraneous -

The high temp alarm was from the Icy Dock drive cage for Disk 7 (A Seagate 1.5TB).  Cage is set to alarm at 55 degrees, but unRAID was showing it at the high 40s.  So I just told my wife to stop the server then power down.

Had to restart numerous times to get the server to show up on the network.  That's when Disk 1 showed up as Disabled.  I went in to the server hardware, checked cables, powered back up and then started the rebuild.

 

Would a rebuild with a new drive (theoretically) go faster?

syslog-2011-04-13.txt

[kizer]

When I rebuilt my 1.5TB a while back I think it took like 300 or so minutes then the Parity check of 400 or so minutes. Of course those times are off the cuff estimates on a 2.7 Sempron, but it was no way near a few days. Like 8hours complete

 

Answer to your question of would a rebuild of a new drive go faster? Well I would say a rebuild with a properly working system should go faster. Of course system meaning as long as there isn't a true drive error, controller, mother board or various other possibilities.

 

With a new drive you need to run a preclear or you should at least, which as you know will take hours if not days depending on the size. Thats why I keep a spare on hand precleared ready to drop back in to replace either my parity or data disks.

[lionelhutz]

You had some SATA link resets on these devices. Is this an add-in card?

 

Apr  9 20:15:01 Mediaserver emhttp: pci-0000:00:1f.2-scsi-1:0:0:0 host10 (sdl) WDC_WD20EARS-00S8B1_WD-WCAVY1873161

Apr  9 20:15:01 Mediaserver emhttp: pci-0000:00:1f.2-scsi-1:0:1:0 host10 (sdm) ST31500341AS_9VS0PMZT

 

They reset 4 or 5 times within about a 1/2 hour on Sat and then the sdl device dropped to 1.5 Gbps and it appears they began to behave.

 

I also see a apcuspd error connecting to the UPS. There was another user who had server speed trouble when the UPS didn't connect. You should check if it's still running and kill it if it's still running.

 

Did you try to delete a bunch of directories off disk1?

 

Peter

 

 

[Phil C.]

Disk 1 is connected to the onboard SATA port.

I tried to set up apcuspd a while ago, didn't work, thought I had gotten rid of all of it.  Guess not... hmmmm... have to figure out how to delete it.

I did not delete or attempt to delete anything off Disk 1.

Appreciate the help, please keep the ideas coming...

 

[Phil C.]

@kizer

  Okay, now you have me wondering.  Is there any way to run some kind of diagnostic to figure out if my system is working properly?  Of late, my server has been taking a long time to spin up drives, and occasionally video streams will pause to let the buffer catch up.  I have reset my network (router, etc) and all of the other gear on my network is running just fine.  I have set up jumbo frames - problems started after (but not immediately after) doing so.  Can jumbo frames be the problem?  All my gear is rated to handle jumbo frames....

 

@lionelhutz

  I guess the Admiral was doing some maintenance on her iTunes library and iTunes attempted to delete some directories.  That's where that came from.  Now that I look at it (not that I understand it) that syslog looks downright ugly.