Veritas Disks Path Question

From: Carlos Sevillano (carlos_sevillano@ureach.com)
Date: Mon Dec 06 2004 - 11:34:45 EST


Solaris 8
Veritas 3.2
Emulex/EMC/lpfc 5x

On many of my Veritas installations with two Fiber Paths
sometimes veritas shows the disk twice as two separate paths ei.
 When doing a vxpring on the disks on c2 and on c4 Veritas shows
the disks to be two different disks with one path. However,
from inq and the configuration... I know that they are the same
disks seen on two paths. The normal view for something like
this is one path showing on vxdisk list and when doing vxdisk
list cxtxdxsx it should show TWO paths. Has anyone seen these
configurations?

# vxdisk list
DEVICE TYPE DISK GROUP STATUS
c0t0d0s2 sliced - - online
c0t1d0s2 sliced - - online
c0t2d0s2 sliced - - error
c0t3d0s2 sliced - - error
c2t0d157s2 sliced - - online
c2t0d158s2 sliced - - online
c2t0d159s2 sliced - - online
c2t0d217s2 sliced - - error
c2t1d15s2 sliced - - online
c4t0d157s2 sliced disk03 rootdg online
c4t0d158s2 sliced disk04 rootdg online
c4t0d159s2 sliced disk05 rootdg online
c4t0d217s2 sliced - - error
c4t1d15s2 sliced disk06 rootdg online

For each disk on c4 I get different information (as if they are
two different disks):

# vxdisk list c2t0d157s2
Device: c2t0d157s2
devicetag: c2t0d157
type: sliced
hostid: sss01
disk: name= id=1059582980.1784.sss01
group: name=rootdg id=1000507165.1025.sss01
flags: online ready private autoconfig autoimport
pubpaths: block=/dev/vx/dmp/c2t0d157s4
char=/dev/vx/rdmp/c2t0d157s4
privpaths: block=/dev/vx/dmp/c2t0d157s3
char=/dev/vx/rdmp/c2t0d157s3
version: 2.1
iosize: min=512 (bytes) max=256 (blocks)
public: slice=4 offset=0 len=26508480
private: slice=3 offset=1 len=2879
update: time=1100829039 seqno=0.25
headers: 0 248
configs: count=1 len=2104
logs: count=1 len=318
Defined regions:
 config priv 000017-000247[000231]: copy=01 offset=000000
enabled
 config priv 000249-002121[001873]: copy=01 offset=000231
enabled
 log priv 002122-002439[000318]: copy=01 offset=000000
enabled
Multipathing information:
numpaths: 1
c2t0d157s2 state=enabled

The other path:

# vxdisk list c4t0d157s2
Device: c4t0d157s2
devicetag: c4t0d157
type: sliced
hostid: sss01
disk: name=disk03 id=1059582980.1784.sss01
group: name=rootdg id=1000507165.1025.sss01
flags: online ready private autoconfig autoimport imported
pubpaths: block=/dev/vx/dmp/c4t0d157s4
char=/dev/vx/rdmp/c4t0d157s4
privpaths: block=/dev/vx/dmp/c4t0d157s3
char=/dev/vx/rdmp/c4t0d157s3
version: 2.1
iosize: min=512 (bytes) max=256 (blocks)
public: slice=4 offset=0 len=26508480
private: slice=3 offset=1 len=2879
update: time=1100829039 seqno=0.25
headers: 0 248
configs: count=1 len=2104
logs: count=1 len=318
Defined regions:
 config priv 000017-000247[000231]: copy=01 offset=000000
enabled
 config priv 000249-002121[001873]: copy=01 offset=000231
enabled
 log priv 002122-002439[000318]: copy=01 offset=000000
enabled
Multipathing information:
numpaths: 1
c4t0d157s2 state=enabled

I have not done extensive testing with removing the path on the
1 in 10 machines that have this configuration. I seen the
configuration even after doing a full re-install of the systems
(new OS, new driver etc). I believe that if the disks are init
under C4 and I remove the C4 Fiber Path from the system to
simulate a Fiber card failure, that I will loose path to the
storage and would need to reboot to come up on c2. I know that
is how it works with Veritas from before I started using
PowerPath.

I am not too worried... since I installed power-path on top of
the 1 in 10 machines that show that way... I disabled either
path and survived on the other. However, I did not do the tests
with just veritas... again I believe that with just veritas I
would have to reboot to come on the other path. Several times I
lost a path and that is what I had to do (Veritas did not
survived online, a reboot was required).

Does anyone know why one some systems Veritas Shows two
different paths for the same disk?
Has anyone seen that configuration and done testing on it? What
makes 1 in 10 machines show the paths this way?

Carlos

ps normally a machines working correctly looks like this (Most
of my systems look this way; and when doing a vxdisk list
vxtxdxsx it shows two paths to the same disk):

#vxdisk list
DEVICE TYPE DISK GROUP STATUS
c0t3d0s2 sliced rootdisk rootdg online
c1t3d0s2 sliced rootmirror rootdg online
c4t1d135s2 sliced localdg07 localdg online
c4t1d136s2 sliced localdg08 localdg online
c4t1d137s2 sliced localdg09 localdg online
c4t1d144s2 sliced localdg10 localdg online
c4t1d145s2 sliced localdg11 localdg online
c4t1d146s2 sliced disk02 rootdg online
c4t2d7s2 sliced localdg04 localdg online
c4t2d8s2 sliced localdg05 localdg online
c4t2d9s2 sliced localdg06 localdg online
c4t2d10s2 sliced localdg01 localdg online
c4t2d11s2 sliced localdg02 localdg online
c4t2d12s2 sliced localdg03 localdg online
c5t2d7s2 sliced - - online
c5t2d8s2 sliced - - online
c5t2d9s2 sliced - - online
c5t2d10s2 sliced - - online
c5t2d11s2 sliced - - online
c5t2d12s2 sliced - - online

#vxdisk list c4t1d135s2
Device: c4t1d135s2
devicetag: c4t1d135
type: sliced
hostid: pppssp01
disk: name=localdg07 id=1059499921.1212.pppssp01
group: name=localdg id=1004989801.1062.pppssp01
flags: online ready private autoconfig autoimport imported
pubpaths: block=/dev/vx/dmp/c4t1d135s4
char=/dev/vx/rdmp/c4t1d135s4
privpaths: block=/dev/vx/dmp/c4t1d135s3
char=/dev/vx/rdmp/c4t1d135s3
version: 2.1
iosize: min=512 (bytes) max=256 (blocks)
public: slice=4 offset=0 len=26508480
private: slice=3 offset=1 len=2879
update: time=1098590547 seqno=0.18
headers: 0 248
configs: count=1 len=2104
logs: count=1 len=318
Defined regions:
 config priv 000017-000247[000231]: copy=01 offset=000000
enabled
 config priv 000249-002121[001873]: copy=01 offset=000231
enabled
 log priv 002122-002439[000318]: copy=01 offset=000000
enabled
Multipathing information:
numpaths: 2
c4t1d135s2 state=enabled
c5t1d135s2 state=enabled

________________________________________________
Get your own "800" number
Voicemail, fax, email, and a lot more
http://www.ureach.com/reg/tag
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers



This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:29:50 EDT