NASLite Network Attached Storage

I've been having problems using an array I just built with NASLite 2HDD and I'm wondering if it's a caching issue with NASLite, if I'm running out of memory or what.

Anyway, I'm currently running a 2TB Raid 5 array (using a Megaraid 150-4 3 1TB WD GP RE drives) on a machine w\512mb ram.

I've run the Raidcard diagnostics on the drives in question and no problems were found, I ran Memtest and no memory issues showed up there.

I get the follow (a small sample of the entire log) when I try to copy more than a couple hundred gig to the Raid array (though the 500gb non-raid drive I've got on there seems to not have any issues).

Am I simply running out of caching space and is there anyway to fix this? The machine is maxed out at 512 with as much as it is specified to handle and I don't have a newer machine that I can use this with.

Log Sample:
Jun 16 01:40:07 user.err kernel: 0(3,1)
Jun 16 01:40:07 user.err kernel: ext3: No journal on filesystem on sd(8,1)
Jun 16 01:42:18 user.crit kernel: EXT2-fs error (device sd(8,1)): ext2_free_blocks: Freeing block in system zone - block = 16777216
Jun 16 01:44:04 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:44:04 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 01:46:59 user.crit kernel: EXT2-fs error (device sd(8,1)): ext2_free_blocks: Freeing block in system zone - block = 16777216
...
...
more of similar
...
...
Jun 16 04:43:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 04:43:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 04:44:06 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 04:49:33 user.crit kernel: EXT2-fs error (device sd(8,1)): ext2_new_block: Free blocks count corrupted for block group 4574
Jun 16 04:49:33 user.crit kernel: EXT2-fs error (device sd(8,1)): ext2_new_block: Free blocks count corrupted for block group 4574
...
more
...
Jun 16 09:09:02 user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 09:17:03 Pasis_NAS user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 09:20:02 Pasis_NAS user.warn kernel: memory.c:100: bad pmd 01000000.
Jun 16 09:20:02 user.warn kernel: memory.c:100: bad pmd 01000000.
...

I don't even know why it came up with "Pasis_NAS" considering the actual name is "Paris_NAS" and everything else shows the correct name (other computers, etc.)

Hoes the problem occur during mirroring only or does it also happen during normal client reads and writes? Some hardware info will also be helpful - perhaps a syslog. I haven't seen this reported before so I'd say it's probably a hardware issue.

The issue occurs whether using windows to do a copy of the files or to transfer files via FTP to the Raid-5 array and occurs after transferring around 100GB of data. I originally used windows 'copy' to copy files from one server drive to the second, had a memory failure, went through the arduous process of verifying the Raid-5 file system (and used the MegaRaid Bios the check drive consistency) and then switched to trying FTP as it seemed to be consistently faster and give better log output. The same thing has now occurred.

After my original post I went ahead and ran memtest again overnight and got results different from the last time I ran it (just before posting my original message) and these included some errors on one of the sticks of memory.

This leads me to hope that the problem is being caused by the one stick of memory.

I'm running memtest just on the second stick right now and will be getting more memory some time soon and test that as well to see if it was the memory, sockets, a combination of either, etc.

Once I verify consistent test results with memtest I'll bring the server back up with NASLite, run the file system check and try copying the data again.

I will let you know the results of the testing when they are done.

When I bring NASLite up again I can get a copy of the hardware/driver info from the syslog, I left it out for the sake of space.

If memtest produces different results each time you run it, then the problem is definitely related to RAM. It is possible that temperature changes cause RAM to falter, so the problem appears to be intermittent. I'd leave memtest running for a few cycles to make sure that each stick is in fact good. My bet is that you'll end up logging errors at some point, indicating that you do in fact have bad RAM.