anybody having *MANY* fs errors?

[pbharris]

hi all,
i recently moved to kernel 2.4 (i am using wolverine) and i am haing many file system errors? i checked the drive completely sing the vendors dianostic tools and it says everything is okay. Here is my setup:

Tbird 800 MHz on an Asus A7V MB.
the drive in question is connected to a promise UDMA controller on board the MB. My hdparms looks like this:
/sbin/hdparm -c 1 -d 1 -k 1 -S 180 /dev/hde

I used kernel 2.2.18 with it but everything was located on /dev/hda , with that drive only being used for storage. i have tried upgrading to kernel 2.4.3 but i ma getitng errors in the file all the time even though the gpg sign is checking out.

any ideas???

[Lorithar]

*ummms*


What sort of filesystem errors are you getting ...

the HD parms look okay ... you note that the manufacturers program is vetting the drive... does it run in linux and does it run across the same controller when it does this?
(most manufacturers validation programs run in windows... )

Last time I experienced this sort of problem it was the io controller on the motherboard decomposing ..(grin) it eventually (about an hour later) died horribly...

[pbharris]

No, the vendor tool does not use the same controller, it only sees /dev/hda (i need to hook up the drive to the main IDE channel). it uses a dos boot disk. in fact it did just die. fortunalty i have a nice debian box.


kinds of errors???? all of them i think, misplaced inodes, lost clusters, wrong count on something, it seems to go to crap the longer the box is up. e.g. leaving box up for one day will give fewer errors than when i leave it up for 5 days.... the memory checks out okay, soe doe the processor.

[MkIII_Supra]

I have noticed the opposite, the longer I leave my 2.4 series systems up the better they seem to run, I have to kill X once in a while just to refresh it, but other than that I have had very few problems with my system.

I too am running Wolverine. Love it, everything works including VMware! But I am also running a test kernel, 2.4.1-0.1.9, the 2.4.2 series wouldn't work with my gateway, it kept deleting my 3COM and nothing I tried would get it back.

I am very pleased with Red Hat so far. Now if I only had the ReiserFS then I would be really happy! Wolverine doesn't support it.... yet.

[pbharris]

thanks people. i think i will try it again... perhaps use that promise patch....

[Lorithar]

*hmms*

The UDMA 100 controller on that mobo may well be dying ... keep your eyes on the disk performance ... if it's degrading slowly and producing more and more errors ... be prepared to loose the controller... trust me onthis ... One point .. make very sure that you are using an ATA100 capable cable on that drive, and keep a close eye on the drive performance. I've seen IDE controllers die slow horrible deaths ... leaving behind perfectly fine disks, but because the controller is dying it reports dozens of bad sectors, failed read/writes, filesystem checks.

Since you point out that you move the drive to verify it (from the accellerated controller to the regular IDE controller), this CONFIRMS to me that the controller is just about pooched.

For the sake of your sanity and data DO NOT use any corrective measures on the drive whilst it is connected to the accellerated controller.

You mention patches ... which patches have you not applied to the driver?

[pbharris]

hey Lorithar,
promise has has a patch for the kernel at their web site, but it is also included with the stock kernels and trying to apply the patch was kinda worthless because i was not able to finish compiling the kenrel.

i am not sure if it is the controller of the disk utility simply could not see the UDMA100 controller, so i could not do anything there. i finally decided to move it to hdb.

i do indeed get more errors over time.
oh well, i don't think UDMA100 is a big improvemnet over UDMA66 anyway :) I did a low level format of the drive to make sure there were no errors residing on it and then reinstalled, i hope my backups are okay...

[jkm]

Don't press the power button, you have to UNmount first. :p

[mike32]

the A7V uses the VIA KT133 chipset which has has corrupption problems with 2.4.0/1/maybe 2. I have a K7V with KX133 chipset with no problems with wolverine with the kernel it comes with. If i was u i'd wait till the next beta or final od rh which will prberly be out very soon i'd say which has the 2.4.2 kernel which fixes the problems with the KT133 chipset.

Mike