When using VxVM Campus cluster feature, volume plexes are automatically attached when a site is manually detached from a disk group

Description

Error Message

None

Cause

The vxattachd daemon monitors vxvm for events and attempts to proactively correct the VxVM configuration. In this case, the daemon reacts to the disk event by attempting to reattach and detached plexes. It does not take into account that the plexes were manually detached due to administrator intervention.

Resolution

A fix in the form of a hotfix for 6.0.5 and 6.1 is available, please contact support for the hotfix.

If a site is manually detached for a short period, for example to carry out planned maintenence at that site, then a workaround is to kill the vxattachd deamon prior to detaching the site. Then once the site has been reattached following completion of the maintenence window the vxattachd daemon can be restarted.

To stop the daemon, locate vxattachd in the process table and kill it, for example on Linux:

pkill -9 `pgrep vxattachd`

To restart vxattachd:

vxattachd root &

See the man page for vxattachd for more information.

Applies To

This issue effects all VxVM versions and platforms

Issue/Introduction

When using VxVM cmpus cluster with site consistency enabled with 2 sites, 'site_a' and 'site_b'. The resulting layout is as follows:

dg testdg default default 20000 1396882014.20.server101
sr site_a ACTIVE

sr site_b ACTIVE

dm d1 ibm_shark0_2 auto 65536 2027264 -

dm d2 ibm_shark0_3 auto 65536 2027264 -
v vol01 - ENABLED ACTIVE 204800 SITEREAD - fsgen

pl vol01-01 vol01 ENABLED ACTIVE 204800 CONCAT - RW

sd d2-01 vol01-01 d2 0 204800 0 ibm_shark0_3 ENA

pl vol01-02 vol01 ENABLED ACTIVE 204800 CONCAT - RW

sd d1-01 vol01-02 d1 0 204800 0 ibm_shark0_2 ENA
dc vol01_dco vol01 vol01_dcl

v vol01_dcl - ENABLED ACTIVE 67840 SITEREAD - gen

pl vol01_dcl-01 vol01_dcl ENABLED ACTIVE 67840 CONCAT - RW

sd d2-02 vol01_dcl-01 d2 204800 67840 0 ibm_shark0_3 ENA

pl vol01_dcl-02 vol01_dcl ENABLED ACTIVE 67840 CONCAT - RW

sd d1-02 vol01_dcl-02 d1 204800 67840 0 ibm_shark0_2 ENA

Manually detach 'site_b', the plex at site_b is detached and the site changes to 'OFFLINE' state:

 vxdg -g testdg detachsite site_b

dg testdg       default      default  20000    1396882014.20.server101
sr site_a       ACTIVE
sr site_b      OFFLINE
dm d1           -            -        -        -        NODEVICE
dm d2           ibm_shark0_3 auto     65536    2027264  -
v  vol01        -            ENABLED  ACTIVE   204800   SITEREAD  -        fsgen
pl vol01-01     vol01        ENABLED  ACTIVE   204800   CONCAT    -        RW
sd d2-01        vol01-01     d2       0        204800   0         ibm_shark0_3 ENA
pl vol01-02     vol01        DISABLED NODEVICE 204800   CONCAT    -        RW
sd d1-01        vol01-02     d1       0        204800   0         -        NDEV
dc vol01_dco    vol01        vol01_dcl
v  vol01_dcl    -            ENABLED  ACTIVE   67840    SITEREAD  -        gen
pl vol01_dcl-01 vol01_dcl    ENABLED  ACTIVE   67840    CONCAT    -        RW
sd d2-02        vol01_dcl-01 d2       204800   67840    0         ibm_shark0_3 ENA
pl vol01_dcl-02 vol01_dcl    DISABLED NODEVICE 67840    CONCAT    -        RW
sd d1-02        vol01_dcl-02 d1       204800   67840    0         -        NDEV

After manually detaching the site, mimic a storage problem and recovery:

# vxdmpadm disable dmpnodename=ibm_shark0_2
# vxdmpadm enable dmpnodename=ibm_shark0_2

vxattachd automatically reattaches the plex back to the volume. The site remains in 'OFFLINE' state:

[root@server101 ~]# tail -f /var/log/messages
Apr  8 01:56:14 server101 kernel: VxVM vxdmp V-5-0-148 [Info] enabled path 8/0x80 belonging to the dmpnode 201/0x40
Apr  8 01:56:14 server101 kernel:
Apr  8 01:59:14 server101 vxvm/vxplex[3618]: Volume vol01 of tonydg: Using FMR for plex attach


dg testdg       default      default  20000    1396882014.20.server101

sr site_a       ACTIVE
sr site_b      OFFLINE	
dm d1           ibm_shark0_2 auto     65536    2027264  -
dm d2           ibm_shark0_3 auto     65536    2027264  -
v  vol01        -            ENABLED  ACTIVE   204800   SITEREAD  -        fsgen
pl vol01-01     vol01        ENABLED  ACTIVE   204800   CONCAT    -        RW
sd d2-01        vol01-01     d2       0        204800   0         ibm_shark0_3 ENA
pl vol01-02     vol01        ENABLED  ACTIVE   204800   CONCAT    -        RW
sd d1-01        vol01-02     d1       0        204800   0         ibm_shark0_2 ENA
dc vol01_dco    vol01        vol01_dcl
v  vol01_dcl    -            ENABLED  ACTIVE   67840    SITEREAD  -        gen
pl vol01_dcl-01 vol01_dcl    ENABLED  ACTIVE   67840    CONCAT    -        RW
sd d2-02        vol01_dcl-01 d2       204800   67840    0         ibm_shark0_3 ENA
pl vol01_dcl-02 vol01_dcl    ENABLED  ACTIVE   67840    CONCAT    -        RW
sd d1-02        vol01_dcl-02 d1       204800   67840    0         ibm_shark0_2 ENA

Welcome to "KB Articles"