Intermittent Saving and Disk Errors with SMB Share
Posted: July 25th, 2014, 11:49 am
To provide some background, I am running unraid in a VM that hosts a temp share (with rw permissions for anyone on my network) via smb and nfs. I have an ubuntu 14.04 server vm that runs sabnzbd, and mounts the unraid share using smb (via fstab) at boot.
Most of the time it works just fine. I can read and write, and no errors are reported by unraid. However, if I start a bunch of downloads up, I will intermittently get saving errors and disk errors. Sometimes the time between them will be nearly immediate, while other times, it can be several hours of downloading until it pops up again. Moreover, these disk errors then tend to correlate with HUGE sysloads up to 12.50 (despite allocating 6 cpus, 8gb of ram, setting the article cache to -1 and reducing the number of connections to 20). This does not always happen though (I have 5 errors right now with .16 sysload).
The errors come in the following forms:
"Disk error on creating file /mnt/temp/downloads/incomplete/theNameofTheFileBeingDownloaded/filename.r03"
"Saving /mnt/temp/downloads/incomplete/theNameofTheFileBeingDownloaded/__ADMIN__/SABnzbd_nzo_G4aPch failed"
"Failed to remove nzo from postproc queue (id) SABnzbd_nzo_LmM2wk"
I have tested all the drives with badblocks and unraid's preclear (and two of them are rather new wd reds), so I really doubt there are actual disk errors. Unraid also reports no errors. This has also persisted across three reformats of the ubuntu vm (two paravirtualized, and the last is hardware virtualized), and two reformats of the unraid vm (one paravirtualized, and the other hardware).
Any ideas of what to look into next?
Most of the time it works just fine. I can read and write, and no errors are reported by unraid. However, if I start a bunch of downloads up, I will intermittently get saving errors and disk errors. Sometimes the time between them will be nearly immediate, while other times, it can be several hours of downloading until it pops up again. Moreover, these disk errors then tend to correlate with HUGE sysloads up to 12.50 (despite allocating 6 cpus, 8gb of ram, setting the article cache to -1 and reducing the number of connections to 20). This does not always happen though (I have 5 errors right now with .16 sysload).
The errors come in the following forms:
"Disk error on creating file /mnt/temp/downloads/incomplete/theNameofTheFileBeingDownloaded/filename.r03"
"Saving /mnt/temp/downloads/incomplete/theNameofTheFileBeingDownloaded/__ADMIN__/SABnzbd_nzo_G4aPch failed"
"Failed to remove nzo from postproc queue (id) SABnzbd_nzo_LmM2wk"
I have tested all the drives with badblocks and unraid's preclear (and two of them are rather new wd reds), so I really doubt there are actual disk errors. Unraid also reports no errors. This has also persisted across three reformats of the ubuntu vm (two paravirtualized, and the last is hardware virtualized), and two reformats of the unraid vm (one paravirtualized, and the other hardware).
Any ideas of what to look into next?