read() retries on a faulty SCSI disk
- From: "siddharthtuli@xxxxxxxxx" <siddharthtuli@xxxxxxxxx>
- Date: 13 Jun 2006 05:50:41 -0700
Hi
I am curious about knowing how scsi disk drivers behave with a faulty
disk. I am having a faulty disk and "read()" system call on that disk
will cause 45 retries (about 10 - 15 minutes) before it return with a
failure.
Nevertheless, I got to know from the error policy of SCSI services
stack for HP-UX pre 11.31 release, that this is the expected behavior
of read() on a faulty disk, ie read() retries 45 times before reporting
error, and that pass-through IO's are not retried. ( Am I correct ?). I
am working on 11.11 and 11.23 OS.
I am eager to know an alternative which I can use instead of read(), to
get notified of error condition without a delay. I cannot use "dd" to
identify this as it hangs on a bad disk making infinite retries.
Please oblige by providing pointers to the same.
.
- Prev by Date: Re: how to check "lun migration" on hp
- Next by Date: Re: how to check "lun migration" on hp
- Previous by thread: how to check "lun migration" on hp
- Next by thread: PAM?
- Index(es):
Relevant Pages
|
|
