HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
Operating System - OpenVMS
cancel
Showing results for 
Search instead for 
Did you mean: 

Slow response time after upgerading to OpenVMS 7.3-2

 
SOLVED
Go to solution
STOCK-EXCHANGE
Occasional Contributor

Slow response time after upgerading to OpenVMS 7.3-2

After upgrading to OpenVMS 7.3-2 from OpenVMS 7.3-1 we start to see performance problem.
Checking the DEC-PS we saw that the response time for some disks is slower then before.
We do have fibre channel disks.
any idea why ?
9 REPLIES
Galen Tackett
Valued Contributor

Re: Slow response time after upgerading to OpenVMS 7.3-2

Have you installed either the latest XFC ECO, or UPDATE ECO (which includes the XFC kit)?

This has some fixes for the XFC which can affect performance.

See this link for more information:

http://www1.itrc.hp.com/service/patch/patchDetail.do?BC=patch.breadcrumb.main|patch.breadcrumb.search|&patchid=VMS732_XFC-V0100&context=openvms:alpha:7.3-2


or use this equivalent TinyURL link:

http://tinyurl.com/6o3za
STOCK-EXCHANGE
Occasional Contributor

Re: Slow response time after upgerading to OpenVMS 7.3-2

Yes, I have installed VMS732_Update-V0300 patch!
Robert Gezelter
Honored Contributor

Re: Slow response time after upgerading to OpenVMS 7.3-2

STOCK-EXCHANGE,

A thought. I don't have the documentation handy, but Fibre Channel/SANs is one of the areas where technology is rapidly evolving and improvements are being made.

One though is to check if the bottleneck is really the disks, or the paths to the disks, I recall that there were some improvements in the path handling, but offhand I must admit that I do not have the details. Carefully check release notes.

Also check if some of your tuning was undone by the upgrade. OpenVMS is generally good about keeping things correct, but I have seen cases where a change was not propogated into the right place and was "lost" on an AUTOGEN.

- Bob Gezelter, http://www.rlgsc.com
Jiri_5
Frequent Advisor

Re: Slow response time after upgerading to OpenVMS 7.3-2

Have you installed patch FIBRE_SCSI V4.0?

However If you have the VMS732_FIBRE_SCSI V4.0 remedial kit installed you should not run the SYS$ETC:FIBRE_SCAN utility. This problem will be addressed in a future OpenVMS V7.3-2 remedial patch kit.
Garry Fruth
Trusted Contributor
Solution

Re: Slow response time after upgerading to OpenVMS 7.3-2

As a consequence of a reboot required by the upgrade, the fibrechannel path selected on multipath disks may have changed, causing some channels to be more heavily utilized than before, and hence affecting performance. From the attached link, it does not seem that path selection algorithm changed between 7.3-1 & 7.3-2. From experience, I can tell you that the algorithm does not produce consistant results. If you did a cluster reboot and changed the order in which you mounted your disks, then those disks would likely have a different path selected.

http://h71000.www7.hp.com/doc/732FINAL/6318/6318pro_007.html#path_select_h

Although unrelated to the upgrade, you may also want to consider ways of maximizing the cache hit rate in the controllers. In a clustered environment, force all cluster members to use the same path to a disk. This way, if two nodes read the same block, one will do cause the disk i/o, the other will benefit from a cache hit. If you are using host-based shadowing, adjust the read-cost of one of the members to be higher than the other members. This way, if a process reads block 1, and then tries to read block 2, the block 2 read would benefit from a cache hit (the read of DSAx would go to the same DGA disk, causing the same controller to be used, and if the controller did some sort of read-ahead or read a larger chunk, then block 2 would be in the cache).
eran_6
Advisor

Re: Slow response time after upgerading to OpenVMS 7.3-2


We have same problem.
2 X GS1280 + 2 X GS140 cluster.
FIBRE_SCSI 4
SYS 6
UPDATE 3
EMC symetrixes , brocade & McData switches.
NO XFC at all.

Common fiber path has much influence on disk io rate.
We use some hand-made proc. to summ
MONITOR DISK IO RATE over all PATHS
and do see IO improvement swiching loaded disk to less loaded path.
Today we are much more disk/path optimized than at vms731 , but still IO is a new bottlneck!

What SAN configuration do you have?
eran_6
Advisor

Re: Slow response time after upgerading to OpenVMS 7.3-2

1. IN OUR TEST CLUSTER XFC IS SET ON
2. IN OUR PROD CLUSTER XFC IS SET OFF.

IN BOTH CLUSTERS WE SAW THE 50% REDUCTION IN DEVICE RESPONSE. SO XFC DOES NOT HELP ANYWAY.

AFTER 15 DAYS OF INVESTIGATION AND AFTER WE SAW THE "STOCK EXCHANGE LETTER" WE THING THAT VMS 7.3-2 HAS A BUG !!!

DONT KNOW IF IT IS A DRIVER BUG, OR AN HBA FIRMWARE BUG.

ANYONE WHO FEELS THE SAME DEGREDATION IN DEVICE RESPONSE , PLEASE ANSWER OUT LETTER.

STOCK-EXCHANGE
Occasional Contributor

Re: Slow response time after upgerading to OpenVMS 7.3-2

VISA this is your call - not another call!
STOCK-EXCHANGE
Occasional Contributor

Re: Slow response time after upgerading to OpenVMS 7.3-2

Visa will open their own case