Operating System - HP-UX
1827854 Members
1463 Online
109969 Solutions
New Discussion

depot creation slow after installing new disks

 
Chris Driver
Occasional Contributor

depot creation slow after installing new disks

Hi,

We have replaced the 2 18Gb Ultra-160 scsi hard disks in our J6000 with 2 73Gb Ultra-320 scsi hard disks. We have found that creating a software depot (using swpackage) now takes 35 mins+ where it used to take 7 mins!

It seems to relate some how on the size of the logical volume. We set up a logical volume of 40Gb and the swpackage too 35 mins+. We then did the same thing but on a 9Gb logical volume which took 7mins.

Our J6000 has 3Gb RAM and has HP-UX 11i (TCOE - December 2003 B6821-10034,5,6) installed
3 REPLIES 3
Robert-Jan Goossens
Honored Contributor

Re: depot creation slow after installing new disks

Hi Chris,

Did you compare the large and the small lvol on the same box ?

Would it be possible to install the latest qpacks and hardware patch bundles ?

http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=QUALITYPACK

http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=B3782HWE

Last question :-) wich version of jfs are you using ?

http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=BASEVXFS

Best regards,
Robert-Jan
Steven E. Protter
Exalted Contributor

Re: depot creation slow after installing new disks

Do you have an inconsitency with scsi setup? ioscan -fnC disk

dmesg # look for problems.

If you have various devices of different speeds on the same scsi change you could be slowing things down. Same thing if there is too much i/o on the scsi chain or an Ultrium tape drive or cd-rom mixed on the same chain as disks.

SEP
Steven E Protter
Owner of ISN Corporation
http://isnamerica.com
http://hpuxconsulting.com
Sponsor: http://hpux.ws
Twitter: http://twitter.com/hpuxlinux
Founder http://newdatacloud.com
Mark Greene_1
Honored Contributor

Re: depot creation slow after installing new disks

Did you setup a new volume group for the large lv's, and if so, what's the new PE size? What model controllers do you have? If the PE size is significantly larger than the block size used by the controller and significantly larger than the on-board cache, you may have introduced IO latency.

Have you tried running iostat while creating a depot to verify the rate of disk activity?

mark
the future will be a lot like now, only later