Hello, I have a large disk, formatted ext2 (I know, I should use ext3, but I was testing the ext2 vs. ext3 performance hit). When my machine goes down, it always takes a long time to perform this disk check at bootup. We're talking 30 minutes.

Today, however, it's taking much, much longer. After an hour, it's still only "4.8%" done. I'm planning to just let it sit her for several more hours. How long should this process reasonably take? What are the possible causes of a "freeze"?

What I really want to do is to edit the /etc/fstab file and reset the offending disk to ext3 (it is not the root disk). But I can't get a command prompt.

I'm a bit confused as to what the best strategy might be to go forward.

Thanks!

Edit: I'm running Fedora core 7 on a quad-core x86_64 Intel machine.

You either haven't run 'shutdown' before turning the machine off or were forced to reboot the machine because of a system freeze. The Linux file system is very robust and usually you'll get away with a long file system check. If you are unlucky, however, vital system files may have been damaged. Keeping '/' on a small partition of its own minimizes this risk.
Journaling file systems keep track (a 'journal') of all read / write operations. So even in case of system failure, the status of the system will be preserved and no checks will be run at reboot. So,wait and choose ext3 if you get to that point,if not,I guess you know what you have to do.

Hi alan_ri,

The system is a master node in control of 4 worker nodes (small cluster). It failed last night, and has failed in the past. When I bring it up, this FS check takes a long time, but never this long.

The / partition is indeed on a different disk drive. The drive which is being checked is RAID5.

I suppose I could unplug the RAID disk, try to reboot, and see what happens. Any thoughts on that?

Morgan

Yes,try that,but tell me what is on RAID5 drive.

How big is this drive? Is it locally attached or remote attached (SAN/NAS)? I've seen fsck take anywhere from 5 minutes to several hours, depending on hardware, size and how it's attached.

trickykid -- I believe the drive is at least 300 Gb, perhaps 500 Gb, and as I mentioned, it is RAID 5.

alan_ri -- Should not be any system stuff on the RAID disk; mostly scientific results which I'd hate to lose, but would not be the end of the world if I did.

The progress meter shows "6.4%" done, so it is moving, slowly but surely. At this rate, it will take 2 days! Hopefully it is encountering problems in the early part of the disk read and will speed up as it progresses.

Morgan

The speed of the test is not necesarily going to be constant, it might just jump at some point (though it could also slow down, not probable, but...). I don't want to be an smart-ass here, and by your first post I think you already know it, but ext3 is more robust, and fsck takes a MUCH shorter time on it (it's one of the reasons why ext3 was created in fact).

The only advise that I can give you is to wait, unless you are willing to risk your data. If you reboot while fsck'ing, you might lose everything if the system is left in an inconsistent state. Note, again, that ext2 is not as robust as ext3.

Is the data you mentioned in the /home directory?

Also the slowdown could be caused by a bunch of smaller files. 500GB is rather large. I've seen a 1TB fsck take a day or two before. Most of the time we'd just kill the check or use tune2fs/fstab to change it from not checking upon reboot, etc.

i92guboj,

Your point on the merits of ext3 over ext2 is well taken. As I said, I was experimenting on the speed advantage, determined it was negligible, and was too lazy to change it back.

It turns out that I just let it sit there for a couple hours. Once it got through the bad inodes, the system came up quickly. Phew! I've never seen it take anywhere near that long to check the consistency of that disk.

I have converted back to ext3, and sure as hell will be backing up my data!

Regards,
Morgan

I am glad that all ended ok and in a reasonable amount of time. I have never used ext2 for anything above 40 gb or so, but with that I had enough. It's takes too much to fsck one of those drives.

Even in ext3, it can take a good amount of time depending on what's inside. I have a couple of 500 drives around, formated in ext3, and if there's a power outage they can take around 20-30 minutes to be checked (and maybe fixed). It's damn fast compared to ext2, but still a fair amount of time.

Cheers.