A few days ago, I descibed the problem:
>
>Last week, we got the following message on the console of our 3/160:
>
> vmunix: xd1a: write restore (drive not on cylinder) -- blk #64, abs blk #64
>
>It didn't look serious, so I just ignored it. However, we got the exact same
>message again last night. Does any one know what the 'drive not on cylinder'
>means? This is our configuration (SunOS 4.1):
>
>CPU: 3/160, one Xy7053 disk controller
>xd1: Fujitsu M2351 Eagle, formatted with
>
The 'drive not on cylinder' message essentially means that the controller
checked the cylinder position of the drive heads and found that they weren't
positioned where the controller thought they should be. From the reponses I
received, I gather that there are (at least) two cases when the head position
is checked by the controller:
1) Just before the drive issues a seek command, the controller checks the
current position of the heads (by--I guess--reading the current position
off the disk and comparing the result to the position the controller has
stored) to determine to how far it has to seek and how to get there.
2) After the controller has issued a seek command and the drive has signaled
the controller that the seek has been completed, the controller again
checks the position of the heads to make sure they are correctly positioned.
Hence, a 'drive not on cylinder error' could occur during either of these two
checks. Unfortunately, the causes of such an error can be either nothing to
worry about or something more serious, depending on which check detected
the positioning error.
For the first case, the cause may be harmless. Such an error can occur if the
drive has been idle for a time before a seek was issued, and the heads have
drifted or moved from the last cylinder accessed. This could be the result
of a bad cable, a power glitch, or possibly heat. A drive which is configured
to do random seeks when idle can also cause this error to occur.
For the second case, the cause may be more serious. In this case, the drive
was told to seek to a particular cylinder and missed--not too good. This
could be the result of a bad cable, or worse, a drive or controller going bad.
As far as which was the cause in our case, I'm not sure. The fact that the
message has only appeared twice and both during times of inactivity (although
only ~20 min in one case) leads me to suspect a power glitch. However, the
fact that both messages occured for cylinder 0 (as implied by the abs block #),
and that this is the fisrt time this message has occured in the 3+ years that
we've had the drive leads me to suspect the drive. I don't think its the
controller, since we have two more drives connected to it, neither of which
have shown any problems. The similar cylinder # indicated by the messages
also tends to rule out cable problems, but it's hard to draw conclusions based
on only two cases! In any case, I'll be keeping my eye on the drive.
Thanks to all who responded:
Jeff Nieusma <nieusma@cs.Colorado.EDU>
Doug Neuhauser <doug@seismo.gps.caltech.edu>
Kevin Sheehan <sundev!fletch!kevin@Sun.COM>
Jiri Dvorak <dvorak@iam.unibe.ch>
Mark Bradley <markb@Solbourne.COM>
Ray Smith <rcsmith@anagld.analytics.com>
John Valdes
valdes@geosun.uchicago.edu
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:11 CDT