NASLite Network Attached Storage

Forgive the undescriptive title but it is a bit hard describing the problem...

I am running Naslite-2 CDD on old AMD K6-2 machine with 128mb ram and 3 drives: drive 0 - 300gb, drive 1 - 250gb and drive 2 - 250gb.

I am connecting the the NAS server from a Macbook running OS X Tiger and the connection is NFS.

Most of the time it is running pretty solid, however, every so often I get the NFS mount disconnect. When I try to reconnect, it wont. Usually when the first disconnect occurs, two more disconnect follow of the two other drives. When I try to telnet into the server, I get connection refused. I run the server headless, so there is no way for me to check the output on the screen of exactly what is going on.

I usually restart the server, and then it goes into a long disk check. %50 of the times, the check goes ok and the system goes back to normal and I can re-mount the shares. The other %50 of the times, the server finish the check, but some of the drives are not mounting. (checking the status after telneting into the server). I then try to run the check and repair again on the drive not mounted. I let it run overnight, but it never seem to repair it. The only way I can repair the drive is boot the server with linux liveCD and run the util to repair the drive. Then when I boot Naslite-2 again, all is well.

I am pretty sure there is nothing wrong with the drive. And this is mostly occur when there is activity on the drive, such as copy, or saving a download to the drive.

Anybody else have a similar experience or knows how this might be fixed.

What is also strange is that the check and repair of Naslite-2 does not work, but when using the fsck util from a linux liveCD such as knoppix, it seems to do the job.

Any input would be much appreciated.

How does that manifest itself on your Macbook? I use the syntax in NL's manual to connect (ie. no automount) & if I shutdown the server, Disk-n is still shown in the sidebar.



You mean when you "Go to server" from the Finder, or via Safari http?



I was hoping to run totally headerless, but realise now that it's not totally practical, so temporarily I've got 1 of my KVM inputs going to the NL box.
It may be permanent as the BIOS won't proceed without a keyboard present.

Normally though, I go in via Safari, & click on the Telnet tab in NL's page, which automagically brings up Terminal. Can you not connect via http?



So, you just press off, then on, switch?



Presumably because it wasn't shut down "cleanly"



What does the SMART status say for the drives?



IME the drive is often the culprit sadly.



Very difficult to debug without a KV connection to your NAS.

I dont automount the shares, but rather user the Finder's "Connect to Server" to mount the NFS shares.



Finder





Well, I am running headless, since I keep the machine in a closet. My bios support disable keyboard error, so I dont need keyboard and mouse connected to the machine.



Yes. However, it doesnt just cut off the power, but rather, it sends a shutdown signal. So I believe the system is properly shutting down.



See above. When I setup the system at the beginning with a monitor, I used that alot (using the power button). It does go through a proper shutdown.



SMART is clean on all drives.



I dont think so. The SMART status is clean. And like I said, using a liveCD repairs the disk fine. And this problem occurs pretty random. Sometime it goes strong for couple of weeks. sometimes couple of days.



I am aware of that, but I cannot have a monitor connected to it, other then for setup.

What I do not understand, is why I can repair the drive using fsck util from a liveCD, but the builtin repair does not work.

Maybe the experts could chip in here, but I wasn't aware that it was safe to hardware switch off any modern OS, only via the shutdown etc. software route.





As an example of HD's, a few days ago I set up a new hd for an iBook, but the backup wouldn't copy files to the new hd successfully. It passed SMART ok. I tried partitioning it for windows, & copied files to it seemingly ok, but again, reading them back caused major errors.
A new replacement hd was fine.

Also, coincidentally, as I was about to reply to your post, I'd started to copy a 6.5 Gig' file from this mac to the NAS. After ~ 450 meg's it stopped copying, & just like you I lost the drive!

I'm currently trying to recover it. I started by using Knoppix 4, but it came up with a series of questions, & as I'm not Linux experienced, I decided to try the windows based recovery tools mentioned in another thread.

I was also looking at the NASlite web stat pages, & can't understand why OS X NFS transfers don't show up as such! All that's showing is http from my browser connection. Anyone any ideas?

I'll let you know if I find anything.

Well, like I mentioned above, on my system, pressing the power button does NOT cut off the power, but rather, it sends a system shutdown signal to the OS and it is shutting down properly.

Regarding the similar problem you are having... Try using any liveCD linux distro that will let you boot to a command line. No GUI necessary.
Once you have the prompt, you need to unmount the drive you want to repair, since its probably automatically mounted by linux.
use this:

$ umount /mnt/hda1

After that, you can run the repair utility:

$ fsck -t ext3 -y /dev/hda1

This will run the fsck utility to repair ext3 filesystem.

I just ran this myself 15 minutes ago on the same drive that NasLite failed to repair, and it repaired it with no problems at all. once this was over and it reported the filesystem as clean, I rebooted into NasLite2, and now Im back to normal. (Until the next time this will occur...the clock is ticking).

Good luck.

Yes, mine does too (bios settable delay on some mobo's I think). However, the delay is only ~ 5 sec's, never varying, & when you compare an orderly shutdown from having a gazillion app's & windows open on MacOS (or Windows) that time can be much longer. Do you know what the shutdown signal mechanism is by any chance? I'm thinking that if a controlled power off shutdown is ok, that might be a way to easily setup a UPS system to tell NL that a power fail is imminent?



That looks very helpful, thanks. I'll try it after a few hours sleep

I've downloaded the free version of R-Tools for Linux, running on Win, & it finds only about 25% of files, + it took an hour to scan the hd ( ~ 100 G of files )



Thanks, I think I'll need that!

Oh, any chance you could check if NFS is showing up in the stat's web pages for your Mac transfers? It seems very odd I've only got http showing.

I initially used:

e2fsck /dev/sda1 (from a forum post)

but gave up as it asked too many ?'s I'm not knowledgeable enough to answer (yet).

I tried a couple of commercial Linux rec' tools which list recoverable files & then you can buy to do the actual recovery.

1 or 2 of them actually crashed in XP (512 megs in pc)!

1 said it could do ~ 25%, another ~ 50%.

So, put hd back in NasLite box, firing up Knoppix 4, I tried your commands.

As I could see a disc icon, I was surprised to get an error message that sda1 wasn't mounted anyway, so it looks like Linux presents a disc icon even if the hd is totally alien to it.

fsck came up with several bad inodes, eventually freezing the box!

Reboot, same again, hmmm..............

I've been somewhat sceptical of the assertion that minimal ram is needed, eg., 64 megs.
So I've had 128 megs in this NAS, & until this failure, it's been fine.

So, I've doubled the ram to 256, ran Knoppix again & this time no crash, & after ~ 15 min's (BTW commercial tools took 1 hr to scan hd!) fsck *seems* to have recovered most (all?) of the files, phew!

Very worrying that the commercial tools are so poor.

From this experience I don't think it's realistic to treat the Nas as an appliance. Some knowledge is essential for a repair situation.

So, like you, I'm now wondering why our hd's crashed during a straightforward file copy. I'll leave the 256 Megs of ram in & see what happens. As I've only got this one 120G sata HD in currently, I'm wondering if 256 is going to be enough when I add 3 more hd's........

Next to find out why NFS transfers are not showing up as such.

Thanks again for your help, sorry I can't help any further yet, but obviously I'll try to find out more.

This is soon to be a repeatable problem.

Reading the gigabit speed thread, I decided to do a few tests here.

I've put a new 250G hd in the NAS, disconnected the gigabit switch, & taken the ethernet cable directly from mini to NAS box.

I started a copy of a 1.6Gig file from mini to Nas. Timed it, did the reverse, then on the next copy, the transfer didn't even start!
No communication with Naslite, so over to KVM connection direct. Hit keyboard, rubbish characters on screen.
Reboot, & it's been about 1.5 hours "busy" on this new hd!

I'll give it another hour, then pull the plug & try Knoppix again.

Is your hd sata or pata?

I've got a PCI sata card in this pc, so I'll try to find a pata hd to eliminate any sata related issues.

"rubbish characters on screen"?

A couple of things you can look at are heat and memory. If the CPU overheats or if your RAM is flaky, then you'll have that problem. I'll take the naslite box out of the closet, open it and make damn sure the CPU and everything else is running cool. Run a memory test on it for a while and make sure the RAM is fine too.

Thanks for keeping us posted.

I use PATA drives in my machines since its really old. I dont know why you have issues with the RAM. I have 128mb on my old machine and I dont have a problem.

There is definitely something going wrong here. Maybe the developer can chime in.

When you reboot the NAS and it says its busy for about 30 minutes, its basically running check and repair. But obviously, the built-in repair in NasLite does not seem to do the job.

Thanks for your response.

Like I suspect JohnnyV, I can't stand the noise from these pc's, so mine's actually in the garage, where it's ~ 4 degrees (C, that is! ) so I doubt heat's a problem But a good point nevertheless.

I do agree that it could be ram, so I'll do a ram test as you suggest. It's odd though that it's been working ok with about 30 gig's comprising ~ 15,000 files each night (max size 200 meg's) for a couple of weeks, & only doing speed (or rather, lack of) tests with big files has brought this to light. But again, buffers in ram could be a problem if said ram is suspect.

I did give up on Naslite's "busy" repair on the hd, as the hd seemed to be static.

Interestingly, I booted from Knoppix which did mount the hd ok, so maybe Naslite's repair succeeded but got "stuck".
I did run fsck though, & it did do several passes & come up with errors.

Ok, I'll go see if there's a comprehensive ram test I can run from a USB stick, else I'll need to get a floppy I suppose.

You're welcome, please update us on any progress at your end

I'll do a RAM test as suggested, but this ram has worked fine in a standard win pc installation.

Agreed on the built in repair. It doesn't inspire any confidence & takes forever to carry out. It's actually quicker to find a CD drive, plug it in, boot Knoppix (slow anyway) & run your fsck suggestion!

Maybe it's a stripped down utility to fit in a small footprint?

If you have improperly set voltages or heat dissipation problems, you could have your box in the fridge and it will still overheat.

Naslite does not speak “rubbish”, so what you get on your screen is in all likelihood caused by your hardware.

That could be anything obviously, but in my opinion it’s likely caused by cpu or ram.

Use this to test http://www.memtest86.com/

Agreed!

Except this mobo has been running win for several years without problems, & *no* changes have been made to it for NL

Naturally, I have checked the BIOS for cpu related settings (& everything else actually), several times




I think that goes without saying,

& yes those are the likely suspects, but again, it has been running previously just fine......



Thanks for the URL. I've burned a CD on my mac from the downloaded image & it's been running a while on the NL box. No errors so far, from either caches or DRAM.

I'd let the test run for a few complete cycles. That may take some time depending on the amount of ram you have. If you can confirm the CPU and RAM are fine, then that would be a good point of reference since it eliminates the two primary suspects.

The next thing I’d look at are the IDE cables as well as the actual naslite media and installation. I've had a number of occasions where reprogramming the usb drive or recreating the boot CD has solved strange problems.

There is a possibility that a driver used by naslite does not like your hardware, but that is the least likely reason.

I’m only trying to help