NASLite Network Attached Storage

My ultra-reliable Naslite server (which has been running 24/7 for several months without issue) has just started playing up.

This morning, in the middle of a file transfer to the server, an error message appeared indicating the transfer could not be completed, and all disks on the server became inaccessible. I've re-booted three times now, but it won't come back. On the third boot, I took the following (partial) notes. All was well until the filesystem check:

And that's where it freezes. It doesn't get as far as Disk 6 (the last disc). All suggestions welcome. As I said at the top, this has been a very stable system, and I haven't made any recent changes to the machine or the network.

My guess is bad RAM. Reseat the RAM and run a test. If it passes, try booting the NAS again.

Thanks for the response. Does RAM "go bad"? (presuming it was good to begin with)

Not generally, but it does happen.

He didn't say your RAM was bad. He told you to re-seat the DIMMs and try booting again, there is a difference. Remove them and then re-insert them into the sockets. The chance that the RAM is bad is very unlikely. There is also a chance that the processor even got jarred loose and it may need to be re-seated as well.

After trying that you may still have bad sectors on the drives since in a fall like that under operation the heads might have "Crashed" into the surface of the platters and damaged the data in those areas, maybe the platter/s and even the heads depending on the severity of the fall.

Mike

I think my (very careless) choice of words may have led to a misunderstanding. When I said in my first post that the server "fell over", I simply meant that it stopped working -- i.e. a file transfer stopped in mid-transfer with an error message indicating the transfer could not be completed, and the system has since failed to boot. Apologies if I confused anyone. I have now edited the first post so that it is clear.

I did, in any case, re-seat the RAM, and have re-booted a few more times. Each time I seem to get a bit closer to a successful boot -- last time the filesystem check got as far as the last (seventh) disk before freezing on a kernel error (although two disks still failed the check). I'll keep trying.

Drop back to one known good DIMM and try to boot the box, you may have had a stick go bad on you.

Mike

Curiouser and curiouser...I also started getting "out of memory - system halted" errors at bootup, which did also seem to suggest a RAM issue. So I swapped out the RAM for known good RAM (it's old PC133 DIMMs in a Celeron 700MHz based system -- 2x128MB of RAM -- 7 HDDs (3xIDE, 4xSATA), Seasonic 430w power supply).

And I'm still getting "out of memory - system halted" errors at bootup, and kernel panics if I get as far at the system file checks. I did manage to boot up once and get as far as the admin login. But although I was able to use the administration utility, the disks were still not available on the network, and I could not browse to the status information pages. One disk had failed to mount.

Then when I rebooted from the admin utility, I only got as far as an "out of memory" error.

I really am at a loss now. This is a system that has run fault-free for a long time. Power supply maybe? That has been suggested before in these forums in response to the "out of memory" error. I switched ACPI off and that has not helped.

EDIT: I've also tried burning a new CD from the Naslite .iso file in case that had developed errors. No change.

EDIT: I find that if I disable ACPI, I am less likely to get the out of memory messages -- but I still get the kernel panics at later points in the boot cycle.

Run a memory test using memtest (http://www.memtest.org/). Let it complete a few cycles before assuming RAM and most of your board is operational. If all that works, then try resetting your BIOS to default settings and try booting NASLite without any storage drives. Check the syslog and make sure all is OK. Then start adding the drives.

With things like that the only approach you can take is by using a process of elimination. Not sure if that makes sense, but it will help you isolate the general area where the problem may be originating from.

Hope that helps.

Okay, so I ran Memtest, and I had about 10 errors in the high-stress test (#5) and #8. Otherwise error-free. So I swapped out the memory for a different set -- and still had five or six errors in test #5 and #8.

So here's my logic -- please rip apart if incorrect. Given that the test failed in similar fashion on two different sets of RAM (both of which have been operating without error up to now), that means the RAM alone is not likely to be the problem. It is either:

(a) a compatibility issue -- unlikely as this system has been running without issue for at least a year.
OR
(b) other hardware is failing: motherboard, CPU or power supply (or SATA card?). Much more likely.

My hunch is that the most likely culprit is the motherboard, so I intend re-building the server with a new motherboard when I get a chance. If anyone thinks I'm taking the wrong approach, please say.

EDIT: After the Memtest session, I tried to boot the server using just a single HD (IDE). It started without any problem and is visible on the network. I can transfer files to it. I am now very confused. I'll guess I'll add back a disc at a time until it fails again...

EDIT: A few hours later, the (single disc) server vanished from the network again. So unless anyone has a better suggestion, I'll press on with replacing the motherboard...

Reading this through, it seems more and more likely that you are having power supply issues. As power supplies get warmer, their capacity reduces, so a cold power supply may boot the machine, but a warm one may drop drives. The fact that you cold booted with one HDD and it later failed would go some way to support this. You have at least 6 drives and presumably a CD-ROM attached - what wattage and make is your power supply?

Seasonic 430 watts.

Is it comparatively new? Capacitor ageing will also reduce output over time.

The power supply is two years old.

Seasonic is a reputable brand, so a two year old unit should be fine (unless this is its second home and it was thrashed to within a milli-amp of its "life" in a previous high-spec PC.....).