Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

varyonvg/varyoffvg command hung?

0 views
Skip to first unread message

rs6000er

unread,
Jan 26, 2006, 2:10:55 PM1/26/06
to
hi all

we are experiencing a strange which can not be resolved by IBM so far.

we are using p5 570 server which is running AIX 5.3 ML03. when I run
command varyonvg/varyoffvg to grab the shared VGs in our FastT, the
process will hang wihtou any response. at the meantime, the
varyonvg/varyoffvg process was not able to be stopoed by kill -9. and
there was not errpt log in host and no error log in fastT.

IBM suggested us to upgrade all firmware for fastT and try again. but
we still got hung after upgrading firmware.


does anybody have the same experience?


Frank

Chris

unread,
Jan 26, 2006, 5:08:08 PM1/26/06
to
Frank,

Need more information....

You said "shared vg." Is this a shared vg for HA failover (only active
on one side or the other) or is this a concurrent vg ?
If it is a concurrent vg.... what app is using it (and you are only
using "raw" devices ... not filesystems, right?)

Additionally, you DID configure the VG with a common Major number,
right?

I'll wait for your response to give you more ideas.

Thanks,

Chris Young
Consultant - Gun for Hire
614-804-UNIX

rs6000er

unread,
Jan 27, 2006, 10:07:23 AM1/27/06
to
Hi Chris

those shared VG are in a HACMP failover tow nodes cluster. they are
regular volume grous. some of them are holding the application
filesystems and some of them are holding the raw partition oracle
database.

i do not understantd what is a common Major number, i just created the
shared volume groups in HACMP and let cluster determined the major
number.

thank you for your response,

any idea will be appreciated,

Frank

Steven

unread,
Jan 28, 2006, 5:24:24 AM1/28/06
to

Frank

Can you explain what you are trying to do again please, I'm not quite
with you.

Are you trying to varyon on one node only? Has the other node also
got the vg varied on? Are these concurrent/enhanced concurrent VG's

Tell us exactly what you are doing.

Dont worry about the common major number, that AFAIK is only needed
for failover. of NFS exported filesystems.

Steven

rs6000er

unread,
Jan 30, 2006, 3:42:23 PM1/30/06
to
Steven

i was trying to varyonvg/varyonvg shared regualr volume groups on tow
clustered nodes. but not at same time like concurrent vg, just varyon
all shared vg and then varyoff all them again on one of two nodes. the
process varyonvg/varyoffvg will get hung if i ran command
varyonvg/varyoffvg on the other node.

i just tried to test if all shared vulume groups are able to be varied
on or varied off on all clustered nodes, node by node.

is it clear?

thanks,

Frank

gertk

unread,
Feb 1, 2006, 3:02:15 PM2/1/06
to
rs6000er wrote:
> Steven

>
> the process varyonvg/varyoffvg will get hung if i ran command
> varyonvg/varyoffvg on the other node.

Do you use multipath I/O with redundant fc paths?
If yes check the status of the RDAC drivers used to connect to the FAStT/DS
and the status of your dacs and dars

fget_config -l dar0
lsattr -El dac0
lsattr -El dac1

The first command will show you on which active paths your hdisks are.
The others will show you the wwn's of the controllers on the FAStT, so
then you can check if all zones are defined ok.

If a LUN has failed over from one controller to the other one, or
zoning/wwn info has changed, this hang behaviour can occur.
What might help is to delete all dac and dar definitions and run cfgmgr
again to get a new proper dac/dar setup

Gert

0 new messages