Skip to contentSkip to conversation

ext3 error with 2.6.9-rc4

DISCUSSION CaT 2 participants 3 messages Oct 12, 2004 – Oct 30, 2004

lore
Share & tools
lore

CaT

author
#0 lore raw

The fs is on a 200gb seagate hd on a promise pci card (20267 - latest firmware). It's hdh1. I was tarring a fs on hde1 onto hdh1. It ran for a bit and then stopped with my kern.log providing the following error:

Oct 13 00:12:03 nessie kernel: EXT3-fs: mounted filesystem with ordered data mode.
Oct 13 00:17:03 nessie kernel: EXT3-fs error (device hdh1): ext3_readdir: bad entry in directory #3522561: rec_len is smaller than minimal - offset=4084, inode=3523431, rec_len=0, name_len=0
Oct 13 00:17:03 nessie kernel: Aborting journal on device hdh1.
Oct 13 00:17:03 nessie kernel: ext3_abort called.
Oct 13 00:17:03 nessie kernel: EXT3-fs error (device hdh1): ext3_journal_start: Detected aborted journal
Oct 13 00:17:03 nessie kernel: Remounting filesystem read-only
Oct 13 00:17:03 nessie kernel: EXT3-fs error (device hdh1) in start_transaction: Journal has aborted
Oct 13 00:17:58 nessie kernel: __journal_remove_journal_head: freeing b_committed_data

A similarish error occured under 2.6.8-rc2:

Sep 19 07:39:29 nessie kernel: attempt to access beyond end of device
Sep 19 07:39:29 nessie kernel: hdh1: rw=1, want=3186822344, limit=390716802
Sep 19 07:39:29 nessie kernel: Aborting journal on device hdh1.
Sep 19 07:39:29 nessie kernel: ext3_abort called.
Sep 19 07:39:29 nessie kernel: EXT3-fs abort (device hdh1): ext3_journal_start: Detected aborted journal
Sep 19 07:39:29 nessie kernel: Remounting filesystem read-only
Sep 19 07:39:29 nessie kernel: EXT3-fs error (device hdh1) in

start_transaction: Journal has aborted

This was during a copy from hde2 to hdh1. 2.6.9-rc4 survived this bit but died anyway when more data was written to the fs when tarring.

The HD is brand new.

Any help I can provide in helping debug this I will gladly give. Just give me a shout.

Signature · 1 line
    Red herrings strewn hither and yon.
#1 lore raw

Hi,

Quoted 6 lines · CaT

The fs is on a 200gb seagate hd on a promise pci card (20267 - latest firmware). It's hdh1. I was tarring a fs on hde1 onto hdh1. It ran for a bit and then stopped with my kern.log providing the following error:

Oct 13 00:12:03 nessie kernel: EXT3-fs: mounted filesystem with ordered data mode.
Oct 13 00:17:03 nessie kernel: EXT3-fs error (device hdh1): ext3_readdir: bad entry in directory #3522561: rec_len is smaller than minimal - offset=4084, inode=3523431, rec_len=0, name_len=0

All this really tells us is that there's something bogus on disk, not how it got there.

There are tools like "dt" which may help identify whether there's data going bad on the way to disk, or whether it might be a fs fault.

http://www.bit-net.com/~rmiller/dt.html

--Stephen

author
#2 lore raw
Quoted 9 lines · Stephen C. Tweedie
Quoted 1 line
Oct 13 00:17:03 nessie kernel: EXT3-fs error (device hdh1): ext3_readdir: bad entry in directory #3522561: rec_len is smaller than minimal - offset=4084, inode=3523431, rec_len=0, name_len=0

All this really tells us is that there's something bogus on disk, not how it got there.

There are tools like "dt" which may help identify whether there's data going bad on the way to disk, or whether it might be a fs fault.

http://www.bit-net.com/~rmiller/dt.html

Thanks for that. A new utlitity to learn. :) Anyways, after getting my laptop (and hence my access to email) back after a week I did a fair bit of testing and the only way I can duplicate the above is by copying from one hd to another. Further testing has led me to believe that the ext3 error is more of a symptom of data corruption caused in the IDE layer somewhere rather then anything else.

I've posted a bug wrt to what I've discovered in the ide update thread (my message before this one).

Signature · 1 line
    Red herrings strewn hither and yon.