VxVM mirror issue

Sameer_Nirmal · ‎12-04-2008

Hi,

We are getting error saying "Operation requires transaction" while trying to mirror a plex of striped volume. In the messages file we see following warning.

vxvm:vxconfigd: [ID 644802 daemon.warning]
V-5-1-11550 Timed out transaction for client 15470. Try setting the environment variable VXVM_TRANS_MAX_TIMEOUT (600) to a higher value and restart vxconfigd

If we re-try the mirroring after few minutes, it works. Does anyone know what would cause it fail and work after some time?

How and where we can set the VXVM_TRANS_MAX_TIMEOUT environment variable? Once set is it safe to

restart vxconfigd?

This host is running Solaris 9 with SF 4.1 and is a part of VCS cluster with EMC storage.

Thanks in advance.

Gaurav_S · ‎12-11-2008

Hi Sam,

This issue occurs because of Veritas Time Finder Utility "Veritas Flashsnap Agent for Symmetrix arrays"

If you are having EMC Symmetrix array & VRTSfas package installed you might this issue.

VRTSfas package is required if you using Business Continuity Volumes (BCV)

Here is a brief about VRTSfas package & impacts if we remove this package.

EMC storage (Symmetrix) has the ability to create Business Continuance Volumes (or internal copies) of LUNs also called Standard Volumes. The copy is a bit-by-bit copy. If the LUNs (Standard) are managed by VERITAS Volume Manager, creating BCV copies will make exact duplicate of Volume Manager objects. If those BCV devices are allowed to be seen by the host where the Standard Devices are also present, Volume Manager is expected to malfunction due to duplicate records and that is not a supported process by VERITAS without VERITAS TimeFinder.

The VERITAS TimeFinder Toolkit and Veritas FlashSnap Agent for Symmetrix allows you to use these copies (BCVs) on the same host as the "original". TimeFinder is a collection of EMC tools for managing the BCV and the copy and split process. VERITAS provides a front-end to the EMC commands (via EMC Application Program Interface - API) and adds extra Volume Manager features and management.

If you remove VxFAS following commands may suffer or may not work.
vxsymsetup, vxsymmir, vxsymsplit, vxsymrestore, vxsymrecover etc.

Hence to summarize if you are using BCV it is suggested NOT to remove VxFAS package.

If VRTSfas package can't be removed following is the workaround to be used to complete the resize operation:
Set a variable VXVM_TRANS_MAX_TIMEOUT=3000

To implement the workaround , we need to do the following:-
(set timeout from default 600 ( 10minutes) to ( 50minutes)

# ksh
# VXVM_TRANS_MAX_TIMEOUT=3000
# export VXVM_TRANS_MAX_TIMEOUT
# env |grep TIMEOUT

VXVM_TRANS_MAX_TIMEOUT=3000

At this stage, we need to restart the vxconfigd daemon ( run vxconfigd -k) so the new timeout parameter is read. Its safe to restart vxconfigd, take appropriate precautions if server is part of cluster...

IF THIS WORKAROUND IS APPLIED VX COMMANDS MAY TAKE CONSIDERABLY HIGH AMOUNT OF TIME TO COMPLETE.

Hope this helps..

Gaurav

Sameer_Nirmal · ‎12-15-2008

Hi Gaurav,

Greatly appreciate your response

We are not using VRTSfas but we do use EMC WideSky TimeFinder Solution Enabler for BCVs. So maybe this product is causing the problem here.

I am curious to know how these agents will cause the problem with VxVM volume mirroring. If this is a known limitation ( I am not aware of it though ), then is there any worksaround for this other than setting the high value for the variable. We don't want to set this variable to high values since it has severe consequence. So this creates one more question to me as under what circumstances or operations with VxVM, we may encounter this issue besides volume mirroring.

Thanks again.

Sam

Message Edited by Sameer Nirmal on 12-15-2008 12:18 PM

Gaurav_S · ‎12-17-2008

Hi Sam,

Unfortunately there is no other discovered workaround on this except setting high timeout value. However I believe this is fixed in 5.x versions.

Besides volume mirroring, vxresize could be another command which might be affected with this.

Sameer_Nirmal · ‎01-19-2009

Hi Gaurav,

Thanks a lot for your response, I do appreciate it. In the same setup, we are now dealing an issue with striped volume relayout. I guess vxassist does use vxresize to do the relayout but not sure.

We get these errors while trying to relayout a striped volume with addition on more ncols.

VxVM vxassist ERROR V-5-1-3119 Operation not valid for <volume> volume
VxVM vxassist ERROR V-5-1-4037 Relayout operation aborted. (20)

This error occurs even after stopping the EMC solution enabler agent.

Maybe it's not related to the vxvm trans timeout variable but wondering what would cause this operation as invalid. I saw one refernence about removing VRTSfas package related to the above message but this package is not installed. We do have VRTSfasdoc package installed on the host though.

Thanks again,

Sam

Gaurav_S · ‎01-20-2009

Hi Sam,

I don't think it relates to timeout value...

What is your current volume structure ? which structure you are relaying out to ?

Is enough space available to move objects ?

Gaurav

Sameer_Nirmal · ‎01-20-2009

Hi Gaurav,

Thanks for your response, I do appreciate it.

The current volume structure is striped and we are trying to relayout it to striped structure

with new disks

There is enough space available. Each new disk has 212113152 and each existing disk has 3072.

I think if it was space issue, it would have given error saying something like "not enough space"

It's weired as it says "Operation is not valid"

Any idea?

Thanks

Sam

Gaurav_S · ‎01-20-2009

Hi Sam,

So if I understand correctly, you are relaying out a stripe volume to stripe volume only, but increasing either stripe width or number of coulmns ?

Also could you share what exact command you are using...

Thanks

Gaurav

Sameer_Nirmal · ‎01-21-2009

Hi Gaurav,

We are increasing the number of columns. The command being used is

# vxassist -g <dg> relayout <volume> layout=stripe ncol=<ncol+4> <disks.......>

springer · ‎01-21-2009

also can we see vxprint -g <dg> -htr volname

the error u are getting usually means there is space isssue

Sameer_Nirmal · ‎01-21-2009

Here is the command output. Please note that we have been trying to relyout using 19 ( 15+4) ncols.

# vxprint -g oradg01 -htr u22

ST NAME STATE DM_CNT SPARE_CNT APPVOL_CNT
DM NAME DEVICE TYPE PRIVLEN PUBLEN STATE
RV NAME RLINK_CNT KSTATE STATE PRIMARY DATAVOLS SRL
RL NAME RVG KSTATE STATE REM_HOST REM_DG
REM_RLNK
CO NAME CACHEVOL KSTATE STATE
VT NAME NVOLUME KSTATE STATE
V NAME RVG/VSET/CO KSTATE STATE LENGTH READPOL
PREFPLEX UTYPE
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT
NCOL/WID MODE
SD NAME PLEX DISK DISKOFFS LENGTH [COL/]OFF DEVICE
MODE
SV NAME PLEX VOLNAME NVOLLAYR LENGTH [COL/]OFF AM/NM
MODE
SC NAME PLEX CACHE DISKOFFS LENGTH [COL/]OFF DEVICE
MODE
DC NAME PARENTVOL LOGVOL
SP NAME SNAPVOL DCO

dm oradg0113 c3t60d29s2 auto 8192 212113152 -
dm oradg0114 c3t60d30s2 auto 8192 212113152 -
dm oradg0115 c3t60d31s2 auto 8192 212113152 -
dm oradg0116 c3t60d32s2 auto 8192 212113152 -
dm oradg0117 c3t60d33s2 auto 8192 212113152 -
dm oradg0118 c3t60d34s2 auto 8192 212113152 -
dm oradg0119 c3t60d35s2 auto 8192 212113152 -
dm oradg0120 c3t60d36s2 auto 8192 212113152 -
dm oradg0121 c3t60d37s2 auto 8192 212113152 -
dm oradg0122 c3t60d38s2 auto 8192 212113152 -
dm oradg0123 c3t60d39s2 auto 8192 212113152 -
dm oradg0124 c3t60d40s2 auto 8192 212113152 -
dm oradg0125 c3t60d41s2 auto 8192 212113152 -
dm oradg0126 c3t60d42s2 auto 8192 212113152 -
dm oradg0127 c3t60d43s2 auto 8192 212113152 -

v u22 - ENABLED ACTIVE 2711617 SELECT u22-01
fsgen
pl u22-01 u22 ENABLED ACTIVE 2764800 STRIPE 15/128
RW
sd oradg0113-15 u22-01 oradg0113 193358592 184320 0/0
c3t60d29 ENA
sd oradg0114-15 u22-01 oradg0114 193358592 184320 1/0
c3t60d30 ENA
sd oradg0115-15 u22-01 oradg0115 193358592 184320 2/0
c3t60d31 ENA
sd oradg0116-15 u22-01 oradg0116 193358592 184320 3/0
c3t60d32 ENA
sd oradg0117-15 u22-01 oradg0117 193358592 184320 4/0
c3t60d33 ENA
sd oradg0118-15 u22-01 oradg0118 193358592 184320 5/0
c3t60d34 ENA
sd oradg0119-15 u22-01 oradg0119 193358592 184320 6/0
c3t60d35 ENA
sd oradg0120-15 u22-01 oradg0120 193358592 184320 7/0
c3t60d36 ENA
sd oradg0121-15 u22-01 oradg0121 193358592 184320 8/0
c3t60d37 ENA
sd oradg0122-15 u22-01 oradg0122 193358592 184320 9/0
c3t60d38 ENA
sd oradg0123-15 u22-01 oradg0123 193358592 184320 10/0
c3t60d39 ENA
sd oradg0124-15 u22-01 oradg0124 193358592 184320 11/0
c3t60d40 ENA
sd oradg0125-15 u22-01 oradg0125 193358592 184320 12/0
c3t60d41 ENA
sd oradg0126-15 u22-01 oradg0126 193358592 184320 13/0
c3t60d42 ENA
sd oradg0127-15 u22-01 oradg0127 193358592 184320 14/0
c3t60d43 ENA

Sameer_Nirmal · ‎05-28-2009

We upgraded to VRTSvxvm 4.1 MP2 with RP4 and still see the relayout problem.

Any idea?

Gaurav_S · ‎06-03-2009

Hi,

Are you still getting same error message ?

From vxprint output above, I see you have 15 disks in the diskgroup & already 15 col stripe exists...

Can you show a vxprint output after adding the 4 new disks ?

Gaurav

VOX

VxVM mirror issue