DMP Veritas with an A/A array (EVA4100) ....

[David Dilly]

Hello i've got a 2 nodes cluster RAC File systeme with an EVA4100.
(OS 11.23 on rx6600 , patch December 2007)
MC-Service Guard : 11.17
HP Serviceguard Cluster File System for RAC : T2777BA

I can see that:

#vxdmpadm getdmpnode nodename=c1t0d2
NAME STATE ENCLR-TYPE PATHS ENBL DSBL ENCLR-NAME
=========================================================================
c1t0d2 ENABLED EVA4K6K 4 4 0 EVA4K6K0

4 Paths for the disk c1t0d2 for the EVA4100 ==> I think it's Good !
Because the EVA It's an Active/Active storage.

#vxdmpadm listenclosure all
ENCLR_NAME ENCLR_TYPE ENCLR_SNO STATUS ARRAY_TYPE
============================================================================
EVA4K6K0 EVA4K6K 50001FE150107540 CONNECTED A/A-A-HP
Disk Disk DISKS CONNECTED Disk


OK th EVA4K6K0 it's really an A/A Array.


But when i exectute this commande :

#vxdisk list c1t0d2
Device: c1t0d2
devicetag: c1t0d2
type: auto
clusterid: cxbanprod
disk: name=c1t0d2 id=1207906025.9.cxorac01
group: name=cfsdg1 id=1207906180.11.cxorac01
info: format=cdsdisk,privoffset=128
flags: online ready private autoconfig shared autoimport imported
pubpaths: block=/dev/vx/dmp/c1t0d2 char=/dev/vx/rdmp/c1t0d2
version: 3.1
iosize: min=512 (bytes) max=1024 (blocks)
public: slice=0 offset=1152 len=10468224 disk_offset=0
private: slice=0 offset=128 len=1024 disk_offset=0
update: time=1207908091 seqno=0.8
ssb: actual_seqno=0.0
headers: 0 120
configs: count=1 len=640
logs: count=1 len=96
Defined regions:
config priv 000024-000119[000096]: copy=01 offset=000000 enabled
config priv 000128-000671[000544]: copy=01 offset=000096 enabled
log priv 000672-000767[000096]: copy=01 offset=000000 enabled
lockrgn priv 000768-000839[000072]: part=00 offset=000000
Multipathing information:
numpaths: 4
c1t0d2 state=enabled type=secondary
c4t0d2 state=enabled type=primary
c6t0d2 state=enabled type=secondary
c8t0d2 state=enabled type=primary

==> We can see 2 path primary and 2 paths secondary .... it's strange for an A/A array no ???

Look again with an other command :

#vxdmpadm getsubpaths dmpnodename=c1t0d2
NAME STATE[A] PATH-TYPE[M] CTLR-NAME ENCLR-TYPE ENCLR-NAME ATTRS
================================================================================
c1t0d2 ENABLED SECONDARY c1 EVA4K6K EVA4K6K0 -
c4t0d2 ENABLED(A) PRIMARY c4 EVA4K6K EVA4K6K0 -
c6t0d2 ENABLED SECONDARY c6 EVA4K6K EVA4K6K0 -
c8t0d2 ENABLED(A) PRIMARY c8 EVA4K6K EVA4K6K0 -

We can see again that the array have 2 paths primary and 2 paths secondary.


I don't understand that because i would have to see 4 paths primary ... NO ???

And how can i do that, to modify this configuration ?

Somebody can help me ?
Thanks.
David

[chris huys_4 · Accepted solution]

Hi David,

Its perfectly normal, what you see there.

The eva4k1, isnt a true and true active/active disk array. (like for example its bigger brother the xp-diskarray)

If im not mistaken the eva4400 has only 2 fc controllers, thus its active/active for this 2 controllers, but each controller can reach the disks through 2 connections, so one of this connections is the active one the other (redundant one) is the passive one.

All reflected in the vxdisk list output.

Greetz,
Chris

[Zinky]

It is correct.

EVA 4x00/6x00/8x00 are "true" active-active arrays just like its big cousins the Hitachi based XP arrays.

I trust you have the Array Support Libraray (ASL) for the EVA4/6/8K Installed too? It is the enabler for VxVM to properly have DMP work with the actove/active nature of these arrays. If not - then you probably have the latest VxVM where it is possible DMP support is already native/built-in and the ASL is already optional.

[David Dilly]

OK thanks for your reply.

I've not installed the ASL, an it work good. I try to failover over the different fabric by disabling the physical port of the switchs and there is no problem.
The DMP of veritas work well !

Thanks again,
David

[David Dilly]

See you soon
Bye