On Wed, 7 Jan 2009, Justin Piszcz wrote:
> Anyone ever seen this before?
>
> I have 4 servers, 2 old and 2 new, all run 6.5.2a, now 6.5.3 as of this
> morning. I see this error occur in the logs a few moments after midnight,
> every night. I did notice mt-st was not installed on the machines that were
> getting this error, so I installed that and we'll see if the issue recurs for
> tomorrow; however, was curious if anyone had seen this?
>
> The first host has two directly-attached fiber drives:
> Jan 4 00:02:19 host1 kernel: st 0:0:0:0: reservation conflict
> Jan 5 00:01:19 host1 kernel: st 0:0:0:0: reservation conflict
> Jan 5 00:01:47 host1 kernel: st 1:0:0:0: reservation conflict
> Jan 6 00:01:18 host1 kernel: st 1:0:0:0: reservation conflict
> Jan 6 00:02:18 host1 kernel: st 0:0:0:0: reservation conflict
> Jan 7 00:01:18 host1 kernel: st 0:0:0:0: reservation conflict
> Jan 7 00:02:20 host1 kernel: st 1:0:0:0: reservation conflict
>
> The second host has one directly-attached fiber drive:
> Jan 1 00:00:45 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 2 00:02:13 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 3 00:00:44 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 4 00:00:45 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 5 00:00:45 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 6 00:00:44 host2 kernel: st 0:0:0:0: reservation conflict
> Jan 7 00:00:44 host2 kernel: st 0:0:0:0: reservation conflict
>
> The two servers are setup identical for the most part as the other two with
> the exception noted above for sotware. For hardware they are using newer
> HBAs etc. However, in 6.5.x I also see the SCSI reservation option has
> changed:
>
> In 6.0MP7:
> Go to Master Server Properties -> Media
>
> [x] Enable SCSI reserve / release
>
> In 6.5.x:
> Go to Master Server Properties -> Media
>
> [x] Enable SCSI reserve
> (o) SPC-2 SCSI reserve
> ( ) SCSI Persistent reserve
>
> 1. Do you think the issue was caused by the absence of the mt-st package on
> the two hosts?
> 2. Or, are the two newer servers (with newer/different qlogic fiber cards),
> do they need a specific settings, perhaps SCSI Persistent reserve and not
> SPC-2 SCSI reserve?
>
> Justin.
>
Minor addition: It is occuring on all four servers, the kernel ring
buffer aka (dmesg) was filled up with other miscellaneous items, I will
change the SCSI reserve to be persistent and see if the errors persist.
Justin.
_______________________________________________
Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|