On Thu, Mar 16, 2017 at 8:47 AM, Peter Sikorski GTL
Hi,
I am getting an intermittent error on my LTO 3 drive. It is a HP StorageWorks 920 internal SCSI drive.
Sometimes backups work and sometimes tar will fail with this error:
tar: /dev/st3: Cannot write: Input/output error tar: Error is not recoverable: exiting now
I have tried 3 old tapes and one brand new tape. It has happened on all of them. I have also run two different cleaning cartridges through the drive.
Also it always fails in a different place. Sometimes it runs for 5 minutes when the error comes up, other times for an hour or so. sometimes not at all and the backup works.
If I look on the monitor connected to the server I see this error:
10:0:3:0 Error on write file mark
A "file mark" is a very low density (or physically large) marker that is written to tape. Tape heads can identify file marks easily while scanning at high-speed, so they often used by tape backup schemes. When you send 1000 files to tape via tar, at the conclusion of writing the entire collection of files, a "file mark" is written, then a "end of tape mark" (or EOT). If you append a second tar backup to the same tape, the EOT will be replaced with the new tar backup, but the first file mark will be left in place. At the end of the second tar backup, a second file mark will be written, and a new EOT. So if you are making your backups with tar, you should logically only get that error at the end of every tar command. Most commercial backup software intersperses "file marks" with the files. The advantage is you can tell the tape drive to advance X file marks and it will do so at high speed. So, I would first try to do a large tar backup. It should run to completion before the error pops out.
Does anyone know what this means exactly or what I should do? I read online that I should try cleaning the tape heads with isopropyl alcohol.
Paul
Once I had a similar problem with an external tape drive. I found out that it was a problem with at power supply not at tape drive. So maybe you should first check cables and connector.
In my experience (and I've done a lot with external tapes), the problem is normally the terminator. Sometimes you can get SCSI to work without a terminator at all, but it will be unreliable. With scsi you need a resistive termination on each end of the cable (ie. you need exactly 2 terminations per connection to the controller, not 1 per device (etc.). Normally the controller is configured to provide the termination at the start of the cable. The termination at the other end is normally a separate physical termination block. Be sure and not just reset those connections, but look at the termination setup and make sure it is right. And yes, it could have worked for years being wrong and now suddenly fail. SCSI termination is reliable when done right, but merely flaky when done wrong.
What’s about the scsii controller? Is only tape connected or also disks? If there are also disks and they work fine, then you can be sure that the controller is OK.
The problem is that from the viewpoint of tar it sends an order to the black box composed of controller, cable, tape drive and tape and get back only “OK” of “fail”. So it’s difficult to find out where the problem really is if you are not so lucky to have spare parts and can change piece by piece… Peter
Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org