At 02:28 PM 1/17/03, Bob Webber wrote:
The obvious first two questions are, (i) Have you checked all the cables; and (b) Were any patches recently applied to the system or devices? Probably worth checking the contents of the mt executable against the same executable on other systems, since the kernel doesn't seem to be logging any actual errors.
How about trussing mt and looking at the return status of the tape ioctls? You might work out more of what's going on by checking the man pages and/or .h files for the meaning of a return which mt only interprets as a generic error I/O error.
I'll second the cable thing, particularly if either the drives or the server use connections for the SCSI in which the plugs are horizontal. I've had lots o' trouble in the past with the stress caused by heavy cables on PCI SCSI cards.
I don't know what your maintenance contract with Overland is but if you have the appropriate level is might be worth talking to them about sending a replacement drive for troubleshooting purposes. The few times I had trouble with Overland drives they were very good about overnighting a replacement drive so it could be checked out.
The other suggestion that I have is to check the usage on the tapes. Yah, yah, I know they're supposed to last rilly long, but I have also seen instances where error rates went up greatly on the "last" recycling of tapes.
One other quick thing. Sit and listen. I don't know this particular model, but on most drives you can tell by the sound whether multiple reread or rewrite attempts are being made. This can be very helpful in separating drive problems from others.
Kids today, they should have used mag tapes when they really sucked.
Joe Yuska
---
Send mail for the `bblisa' mailing list to `[EMAIL PROTECTED]'.
Mail administrative requests to `[EMAIL PROTECTED]'.
