1748070 Members
5277 Online
108758 Solutions
New Discussion юеВ

Re: Mount verification

 
Wim Van den Wyngaert
Honored Contributor

Mount verification

I checked my operator log file and found that sometimes (every few months) "mount verification completed for DSAx" is shown. I checked the activity and found that the disk DSAx had a queue length of 1 and no special load. At the moment (=minute) of the mount verification the queue length goes to 2. Nothin shown in hsg80 consoles.

Why is this happening ? I have it when using set ho /scsi but that was not done at that moment. No anal/dis either. Just DSM activity.

VMS 7.3, GS160 with dual HSG80 shadowed to same config at 10 km.
Wim
9 REPLIES 9
Wim Van den Wyngaert
Honored Contributor

Re: Mount verification

Nothing in error log either.

Wim
Wim
Steve-Thompson
Regular Advisor

Re: Mount verification

Hi Wim

"nothing in the HSG80 consoles..."

Any clues in the the fabric switch log?

Path switching seems the obvious choice, after all!

Steven
Wim Van den Wyngaert
Honored Contributor

Re: Mount verification

No events reported by smtp and nothing in errshow on both switches.

Wim
Wim
Hoff
Honored Contributor

Re: Mount verification

There's not much to go on here. This could be flaky software or flaky hardware or flaky firmware or even (conceivably, very unlikely, but then I've seen weirder) flaky supporting equipment in the computer room.

Any chance you can get to OpenVMS Alpha V7.3-2, or to the current release?

There are great huge piles of stuff that have been updated or have been enhanced since V7.3.

FWIW, a queue length of one is a problem, which implies the software probe is mis-reading or misreporting something in the I/O queue (which has happened; buggy software) or that the disk is heavily loaded.

Longer-term, I'd also be looking at replacing the AlphaServer GS160 and the HSG80 storage with some newer giblets from the AlphaServer GS1280-class (mayhap ES47 or ES80 here?) and with the MSA or EVA-class storage. I'd probably start this sequence by rotating through to newer storage. (No pun intended.)

Regardless of plans to replace the hardware, I'd also confirm the firmware on the HSG80 and on the rest of the I/O stack, and the critical ECOs on OpenVMS. (And again, I'm guessing here.)


Robert Brooks_1
Honored Contributor

Re: Mount verification

Mount verification happens in response to an I/O error that is eligible for mount verification. Certain fatal errors do not result in mount verification. In general, transient disruption in the SAN is a frequent cause of mount verifications that are immediately resolved. If you are only seeing a few of these a month per device, I'd not deem it serious.

On the other hand, I'd STRONGLY urge you to upgrade to V7.3-2. While that version is no longer under standard support, the difference between V7.3 and V7.3-2 for Multipath, the SCSI class driver SYS$DKDRIVER, the fibre channel drivers, and the shadowing driver is pretty staggering!


-- Rob
Steve-Thompson
Regular Advisor

Re: Mount verification

... an interesting scenario,

Just for the record, I've got a cluster with only 1km seperation.

VMS 7.2-2 patched to the gills.
(and thanks to Rob's for his encouragement! :-)

HSG80 - VS86S-10 (Dualled and shadowed)
SANswitch-EL's - V2.6.2
and
ES40's EV68 - FW7.2

It's clustered with Memory channel, the option that works up to 2km. Which is an obvious difference.

I've just checked the operator logs available (2 months) and theres no problems!

???
Steven

Wim Van den Wyngaert
Honored Contributor

Re: Mount verification

No upgrade allowed, no change of hardware will ever be done. Not even installing all patches.

Wim
Wim
Ian Miller.
Honored Contributor

Re: Mount verification

No change means no improvement,

There is a risk in standing still to be balanced against the risk of upgrading.

Did you run FMU on the HSG80 to see if there is anything there? Although I expect that anything there would have been seen on the consoles? (you do have console logging for the HSG80 ?)
____________________
Purely Personal Opinion
Wim Van den Wyngaert
Honored Contributor

Re: Mount verification

As said, nothing in hsg80 console logging. FMU report nothing after 2003.

Wim
Wim