Issue started two days ago. Sabnzbd kept timing out. I changed switches to eliminate that as a cause. I ran a parity check which was successful / no issues detected. Then intermittently the unraid management page stopped loading, unmenu the same. While logging via putty to grab log it I had to try 5 times to get a login prompt. And then sessions disconnected by it self about 45 seconds after copying the log.

 

I can still see all my shares by browsing from a windows pc. I did re-seat the only pci card I have installed, and have pull tried ram in 3 configs (stick 1, slot 1; stick 2, slot 2; stick 1, slot 2) which doesn't seem to be doing anything. I have logged in and edited the go file to stop sabnzb from loading, left everything else in tact.

 

At one point I did pull the USB drive out to copy an old sabnzbd setup onto it, and the copy was going fairly slow. So I thought maybe the drive could be bad. I am not sure if that if that can be the case since it's still serving the shares. Syslog attached.

syslog.txt

You should try upgrading to at least unRAID 4.7. There have been vast changes since unRAID 4.2.

I mis-spoke it was version 4.4.2 I was going off of memory because I couldn't get the http page. I switched the usb drive to a new port and thought that fixed things. I was able to access both the http page as well as the flash drive shave to upgrade to 4.7. I rebooted and things were fine for about 5 minutes, but as  I was on \\tower\flash and clicked on config, I got a network error. It's working now... so seems like it's back to intermittent network errors, with unmenu up I could only see one "yellow" error

 

Apr 28 23:53:51 Tower kernel: Incorrect checksum in table [OEMB] - BD, should be B0 (20090903/tbutils-314) (Minor Issues)

 

I grabbed an updated syslog. The hardware is about 2 years old (feb 09). 

syslog-2011-04-29.txt

Both syslogs look OK, with no real issues.  Sounds like a networking issue, so check some network diagnostics, in particular the ifconfig error and collision counts.  (See the Console Commands for Networking wiki page.)  Perhaps a network card or chipset on either end of the connection is failing, or a cable has been damaged.

Thanks very much for the reply and direction. I'll look into it. (I believe you are probably right, since my last post, I ran memtest and it was successful (well, mostly, after 9 hours without error, I figured... close enough).

I'll mark this closed, I decided to do a shotgun fix and replace mobo/ram/processor. There was a panic stricken 2 days where the network still didn't work and I began to curse the gods, but apparently eth0 is impi and no jumpers can disable it, so I added an eth1 config to my go script.