Anyone have any idea what this means....

[stogma]

Heya..

I have this machine that crashes about once a week. I control it via telnet and I just lose all control of the machine, it won't accept any input. The guys at the remote centre where it is hosted lose all screen access, it goes blank.

So, they do a hard/cold/fast restart by turning the box off and then back on again, as it boots up we have to run e2fsck on it and it gives the following types of errors:

* 'multiple instances of 0 dtime'.
Then it proceeds to request a Y key to proceed with the disk scan.

* 'illegal block found'

It then asks to clear the inode. I don't even know what that means, can anyone explain it to me, i tried searching on redhat but just got pages and pages of rubbish.

* 'inode count is wrong'

It says that it is wrong and then says that it is automatically fixed it, but this happens everytime so it obviously doesn't automatically fix it...

any help is really really appreciated as this is driving me up the wall.

cheers
Stog.

[posterboy]

Hard to do much from there, as the / drive must be unmounted to perform these things. With the drive unmounted, do e2fsck /dev/hdwhateveritis with no options. It will start through the drive in several passes. Each time it suggests a change, accept it, unless you know more about this than I do. After it finishes, type exit, which will force a reboot, after writing the "clean" signature. I have personally never had this fail, but, I know it does. If it won't fix the drive, you are in trouble I cannot help with. Inodes are similar to what the 'Doze world calls sectors. Dtimes are "dirty times" where sectors scheduled to be written, didn't get written because the box went down while they were being held in memory. Linux doesn't write dirty (changed) things at once, rather, it waits until there's a bunch to write together. This info comes from some limited reading I have done, if some of you guys know this stuff better, help us all.The inodes you discard, will go into lost+found on the drive.
Ray


------------------
ray@raymondjones.net
HTTP://www.raymondjones.net

[turdball2k]

Originally posted by stogma:
Heya..

I have this machine that crashes about once a week. I control it via telnet and I just lose all control of the machine, it won't accept any input. The guys at the remote centre where it is hosted lose all screen access, it goes blank.

So, they do a hard/cold/fast restart by turning the box off and then back on again, as it boots up we have to run e2fsck on it and it gives the following types of errors:

* 'multiple instances of 0 dtime'.
Then it proceeds to request a Y key to proceed with the disk scan.

* 'illegal block found'

It then asks to clear the inode. I don't even know what that means, can anyone explain it to me, i tried searching on redhat but just got pages and pages of rubbish.

* 'inode count is wrong'

It says that it is wrong and then says that it is automatically fixed it, but this happens everytime so it obviously doesn't automatically fix it...

any help is really really appreciated as this is driving me up the wall.

cheers
Stog.


Easy one to fix! Put in a Win2k Server CD, reboot, follow the prompts. Be running better than ever in no time... LOL...! (Just kidding). Freaky #coldfusion kiddies...

whaddup st0gma...!

banananachu(nks)