NASLite Network Attached Storage

I've run into my first one of these checks when I pressed the power button on the server, expecting Linux to pick up on this, and shut down the system correctly like other Linux's do, but in NanoNAS's case, this is obviously not the case, and the machine shut off immediately.(at least I know not to do this again! )

It took 27 minutes to check the filesystem on the single 500GB drive in the server so far(and that drive is only half-full), and I have totally lost my patience - Even with Windoze, I'd have the server up and running again inside 3 minutes - even with an illegal shutdown...

I don't consider this filesystem check thing a very clever idea, as with modern drives of several hundred GB each, it is not really practical to have the server offline for eternity while the system checks the disks...

The idea of a server, is to have it up and running again ASAP if there is a power failure or other problem. If NanoNAS needs to run disk scanning, should it not be done AFTER the server is back on line?(such as running Scandisk or CHKDSK /F in XP after XP has restarted) This could be run as a seperate background process after the server is online again...

QUESTION: How do I turn this auto-checking off completely - once I add more drives, I don't want to have to deal with this delay. Once I have 4 drives in there, if NanoNAS has to check each one after an unclean dismount, this would take 1.8 hours to do(based on the 27 minute check-time of the first one), and once I have all 8 drives in place, this would potentially increase to 3.6 hours which is totally ludicrous. If all 8 drives are full to capacity, this could represent 7.2 hours worth of checking time - totally unacceptable for a server.

If I can't totally turn off the auto-checking, I will delete NanoNAS and go back to a Windows 2000-based Server.

In some ways, I hope I have got my maths wrong with respect to the time taken to check(say) 4x 500GB drives as being 1.8 hours, but if I am actually correct, then this is not satisfactory as this is far too long to wait.

Perhaps someone here can give me some pointers?

ACPI support, required for such shutdown, is not included in NanoNAS due to size restrictions.

Okey dokey - fair enough - but can I disable the auto-checking thing to get the server back online again ASAP?

You need to let it finish what it's doing and next time you shut down, use the admin and do a controlled poweroff through telnet. If you power down like you are supposed to, then powering back up will not do a disk check. Problem solved.

Problem not really solved.

While I agree with what you are saying, and even though the server will be running from a UPS, if there is a three hour power cut while I am at work, the UPS will run flat, and we then have the problem.

Generally speaking, our power is pretty reliable, and most power cuts are not that long, but still - with 8 large hard disks, this could still happen, even with the UPS, so I would prefer to disable the checks completely, so that I don't have to come home from work one day, find out there was a four-hour power cut cos some teenage twit crashed his car into a power pole and it took the power company four hours to put up a new one and reconnect the juice.

If I have to wait up to 7 hours for 8 full-to-capacity 500GB drives to be checked due to an unclean dismount, this kinda defeats the purpose of a server at all...(as all this time, it is not serving!)

Do you see the point I am trying to make?

Can one of you confirm that I have my checking-times right or wrong, as I might be incorrect, and if I am then the checking times might not be as drastic as I have indicated they will be...

Yeah, i see your point. It's sort of like wanting your cake and eating it too. I'm not sure you can do that. I don't think nanonas will export a drive that doesn't check out. And i don't think you can disable the check.

Still, if you shut down normally, you don't have to go through the long check time. So what is it that you are worried about? If you turn the nas off like you are supposed to, turning it back on will take little time. THe only time disk check will take a while is if there is an issue with the filesystem. You know that writing to a messed up filesystem can really hose the whole thing for you. Unless you are only reading, in which case you should set the drives to RO so no worries about checks at boot.

Not much else i can tell you.

Great idea, dimension, about setting the drive(s) to RO (Read-Only), because it prevents the check. He still has his "problem", but with some planning most drives will be set to RO, and only one (or two?) set R/W. Plus the RARE nature of the UPS not lasting through a long unplanned power-outage ...

Question: does NanoNAS retain the "poweroff" login found in NL2+ ?

Suggestion to Graeme: look at an automated shutdown script I wrote a couple of years ago ... yes, it requires that an extra WinXP system is up and running ... but it solves your "problem" (even without the RO). Of course, the UPS must have auto-notification (like APC's units do) to the WinXP system it is connected to. Then you can have the WinXP system shutdown the NanoNAS when the notification is triggered (and it shuts itself down thereafter).

http://www.serverelements.com/forums/viewtopic.php?f=8&t=1734
http://www.serverelements.com/forums/viewtopic.php?f=3&t=810

Hope this helps.
Georg

@ georg: That is very, very clever! Had not thought of that - will investigate. I don't have a large UPS, so my concern is that it might run flat before the power comes back on - hense the problem...

@ dimension:
EXACTLY - IF YOU SHUT DOWN NORMALLY.
But if you are not around to shut down normally EVEN WITH THE UPS, you're in trouble.

Example:

0 - I go to work and leave the house.
1 - NanoNAS running happily
2 - There is a power-cut
3 - UPS takes over
4 - Power is not restored anytime soon
5 - UPS runs out of juice and can no longer power the system
6 - System running NanoNAS shuts off with drives still mounted
7 - I come home from work, power is back on again
8 - Drives detected as dirty, and you're stuck in filesystem checks for the next 5 years...

I could fix this problem if I bought a very juicy UPS, but they are very expensive compared to the smaller ones with the likes of a 7aH 12v SLA battery in them.

I suppose I could parallel up several SLA batteries external to the UPS to provide more "On" time in the event of a power-cut...

Can you not set the BIOS to switch the machine back on in the event of power restored? In this way the disk-checking would already be underway before you got home....

Very good point - will check - not sure if the BIOS lets me do that or not, but will check and post back.

please feel free to correct my thinking if i am wrong,

but... i thought the main advantage between a journaled file system (ext3)
and the non-journaled file system (ext2) was exactly this type of problem...
the long file system checks at boot up as a response to an improper shutdown.

it sounds to me that ext2 is the file system, and if journaling was applied to the disks,
this problem would go away then... or would be a non issue since the journal is consulted
during boot up to see if there is any unfinished business.

ext3 takes a lot less time than ext2, but still it take some time.

Should I apply the journal then?
I seem to remember reading either on this thread, or another one that I should leave the filesystem as ext2.

I would love some clarification or advise one way or the other, as it seems there are merits to both...

If I can save even some of that time, then I think the journal would be benificial, however, as the server is going to run 24/7 just like the XP-based one I am running now, I think there was a reason for staying on ext2...

Confused...

apply the journal

Okey dokey - will do.