I have DC++ and uTorrent accessing files on my unRaid server 24/7. Recently I've been getting a lot of 'network path not found' errors on files located on my server from uT and DC just hangs on load when it tries to confirm its hashed files indeed exist in the filesystem.

I was running 4.5.6, I just upgraded to 4.7 - problem persists.

syslog attached.

I *think* the errors are the "Tower kernel: ata9.00: failed command: READ DMA EXT" and "handle_stripe read error: 3035454392/2, count: 1" lines but I'm not sure why or what can be done about it.

dease_syslog.txt

The log shows problems with disk2, /dev/sdj, WDC_WD20EADS-55R6B0_WD-WCAVY1357993. What does the web interface show for disk2?

All disks are green in the UI, though just last week I had to replace a failed disk12

 

EDIT: I didn't notice it all the way over there on the right but disk2 shows 126 errors (even though it IS green), time for a replacement?

All disks are green in the UI, though just last week I had to replace a failed disk12

 

EDIT: I didn't notice it all the way over there on the right but disk2 shows 126 errors (even though it IS green), time for a replacement?

Time for a SMART report on the drive with the errors.

 

Those errors are "read" errors.  Typically they will show as sectors that have been re-allocated, or that are pending re-allocation.  A smart report will tell.    Most of the time, when a read error occurs unRAID will re-construct the un-readable sector from parity and the other data drives, and then also re-write the sector back to the drive where it could not be read.  This "should" allow the disk to re-allocate it.  The SMART report will let you know how many sectors are pending re-allocation and how many were already re-allocated.

 

Joe L.

smart report:

http://pastebin.com/raw.php?i=XY2g2W5e

smart report:

http://pastebin.com/raw.php?i=XY2g2W5e

Your disk has 1445 sectors pending re-allocation and has 5 that have already been re-allocated.

 

As far as I'm concerned, it is time to replace the drive.  It is failing, even though the manufacturer does not seem to be adjusting the VALUE of the parameter affiliated with the "pending-reallocation" downward as the count increases.  To me, that is a bug in their handling of the SMART data for that parameter.

 

Joe L.

 

argh, that sucks, replacing all these drives is getting expensive

what do you guys generally do with drives that have dropped out of an array? I've got 3 WD greens lying around that dropped because of read or write errors but they havent failed "enough" for the WD diagnostics utility to report anything wrong, and from what I understand I can't rma without a diagnostics failure code.

argh, that sucks, replacing all these drives is getting expensive

what do you guys generally do with drives that have dropped out of an array? I've got 3 WD greens lying around that dropped because of read or write errors but they havent failed "enough" for the WD diagnostics utility to report anything wrong, and from what I understand I can't rma without a diagnostics failure code.

 

I have RMAed drives without the code... just pick one that says "can't run WD tool" or whatever.

huh, not sure if I was imagining things last time I tried, but I just went though a very easy rma process without even being asked for a code <3 WD

 

also a big thank you to the regulars for always providing support

You could pre-clear it to death.

huh, not sure if I was imagining things last time I tried, but I just went though a very easy rma process without even being asked for a code <3 WD

 

also a big thank you to the regulars for always providing support

 

FWIW, In the distant past, say 2000, getting a replacement drive from some vendors was a very difficult process at the comsumer level. I found it comparable with getting teeth extracted without novicane. Somewhere along the line, most vendors have changed their process and perception, and, in my experience, made getting replacemnts MUCH easier for us and for them.

 

Bruce