Bad I/O (disk)

[Dave Bunting]

when our Informix Archive reached about 60%, there is an error message "
Archive failed code -1 errno 5".
Reason for failure: Bad I/O on chunk 3.

tried dd:
dd if=/dev/lvolron of=/dev/null bs=4k
dd read error: I/O error
19407+0 records in
19407+0 records out

ioscan ok, pvdisplay, ok, lv and vgdisplays ok diskinfo also ok...
where should the problem be?
Thanks

[Patrick Wallek]

Do your same dd command on all the disks in that logical volume. One of them more than likely has a bad spot on it. When your dd fails on one, call it in to HP to get it replaced.

Hopefully they are mirrored or you have some other form of backup.

[Dave Bunting]

but... how do i prove the disk is really for replacement, if only on LV fails? Does that mean the disk has a bad 'cluster' on the spot the data is being written? Thanks

[Animesh Chakraborty]

Hi,
Which tool are you using for archiving ?
Is it onarchive/onbar/ontape?
archiving to harddisk or tape ?


Thanks
Animesh

[Frank Slootweg]

If a disk fails, then it should be replaced. The 'bad cluster' stuff is from the old (PC) days when disks were rather unreliable and masking out bad sectors/'clusters' was somewhat common (on PCs). 'Modern' disks are so reliable that when they fail they are likely to fail completely very soon, so they should be replaced.

If you have a hardware support contract for the disk in question, then let the support company determine whether or not it needs to be replaced.

[paul courry]

Listen to Patrick. Just do it, because he's right.