Big directory and the UDMA ICRC error writing fsbn message

[Jesse Reynolds]

Hello

Recently, a server of mine running FreeBSD 4.6.2-RELEASE started getting 
UDMA ICRC errors and put it's disk communication into PIO mode. I'm just 
wondering if this could have been caused by one very big and growing 
directory under /var, where one new file was being created every 15 
minutes. It had reached about 100,000 files in the one directory.

Is this possible?

Or is something more prozaic at play here, eg a dodgy cable or a disk on 
the way out?

There are two identical disks on this box that are mirrored with a 
hardware raid chip.

Here are the messages anyhow:

Apr  2 01:13:45 eagle /kernel: ad5: UDMA ICRC error writing fsbn 
119103807 of 0-255 (ad5 bn 119103807; cn 118158 tn 8 sn 39) retrying
Apr  2 01:14:01 eagle /kernel: ad5: UDMA ICRC error writing fsbn 
79975487 of 0-23 (ad5 bn 79975487; cn 79340 tn 12 sn 11) retrying
Apr  2 01:14:02 eagle last message repeated 2 times
Apr  2 01:14:02 eagle /kernel: ad5: UDMA ICRC error writing fsbn 
79975487 of 0-23 (ad5 bn 79975487; cn 79340 tn 12 sn 11) falling back to 
PIO m
ode

I'm also getting some other errors about time, eg:

Apr 15 03:17:13 eagle named[78]: gettimeofday returned bad tv_usec: 
corrected

Apr 18 15:16:54 eagle /kernel: microuptime() went backwards 
(44815198.498490 -> 44815198.-695597409)


about the disks and filesystems:

# df -k
Filesystem  1K-blocks     Used   Avail Capacity  Mounted on
/dev/ar0s1a   1032142    63190  886382     7%    /
/dev/ar0s1g 107383706 94043514 4749496    95%    /large
/dev/ar0s1f   4129310  1432826 2366140    38%    /usr
/dev/ar0s1e   4129310   649442 3149524    17%    /var
procfs              4        4       0   100%    /proc

# atacontrol status ar0
ar0: ATA RAID1 subdisks: ad4 ad5 status: READY


Thanks very much

Jesse