LSM volume shows ENABLED SYNC

[Brendan Murphy_5]

Hi,
I have an LSM volume on a 2 member TruCluster environment, that is constantly showing up as SYNC and ENABLED. The documentation says that this means the plex is sync'ing up. May issue is that this is always the case. What hapens if there is a disk failure? Why is it always trying to SYNC. I never see it as just ACTIVE and ENABLED. Output is shown below. The volume in question belongs to the user1 diskgroup & is called user1_vol.

Thanks in advance for any help.

Brendan


Disk group: rootdg

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg rootdg rootdg - - - - - -

dm dsk2a dsk2a - 21127888 - - - -
dm dsk2g dsk2g - 29995904 - - - -
dm dsk2h dsk2h - 9995904 - - - -
dm dsk8a dsk8a - 21127888 - - - -
dm dsk8g dsk8g - 29995904 - - - -
dm dsk8h dsk8h - 9995904 - - - -

v cluster_rootvol cluroot ENABLED 21127888 - ACTIVE - -
pl cluster_rootvol-01 cluster_rootvol ENABLED 21127888 - ACTIVE - -
sd dsk8a-01 cluster_rootvol-01 ENABLED 21127888 0 - - -
pl cluster_rootvol-02 cluster_rootvol ENABLED 21127888 - ACTIVE - -
sd dsk2a-01 cluster_rootvol-02 ENABLED 21127888 0 - - -

v cluster_usrvol fsgen ENABLED 29995904 - ACTIVE - -
pl cluster_usrvol-01 cluster_usrvol ENABLED 29995904 - ACTIVE - -
sd dsk8g-01 cluster_usrvol-01 ENABLED 29995904 0 - - -
pl cluster_usrvol-02 cluster_usrvol ENABLED 29995904 - ACTIVE - -
sd dsk2g-01 cluster_usrvol-02 ENABLED 29995904 0 - - -

v cluster_varvol fsgen ENABLED 9995904 - ACTIVE - -
pl cluster_varvol-01 cluster_varvol ENABLED 9995904 - ACTIVE - -
sd dsk8h-01 cluster_varvol-01 ENABLED 9995904 0 - - -
pl cluster_varvol-02 cluster_varvol ENABLED 9995904 - ACTIVE - -
sd dsk2h-01 cluster_varvol-02 ENABLED 9995904 0 - - -

Disk group: apps

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg apps apps - - - - - -

dm apps01 dsk7 - 17769412 - - - -
dm apps02 dsk13 - 17769412 - - - -

v apps_vol fsgen ENABLED 17769400 - ACTIVE - -
pl apps_pl1 apps_vol ENABLED 17769400 - ACTIVE - -
sd apps_sd1 apps_pl1 ENABLED 17769400 0 - - -
pl apps_pl2 apps_vol ENABLED 17769400 - ACTIVE - -
sd apps_sd2 apps_pl2 ENABLED 17769400 0 - - -

Disk group: user1

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg user1 user1 - - - - - -

dm user101 dsk6 - 71127888 - - - -
dm user102 dsk12 - 71127888 - - - -

v user1_vol fsgen ENABLED 71127800 - SYNC - -
pl user1_pl1 user1_vol ENABLED 71127800 - ACTIVE - -
sd user1_sd1 user1_pl1 ENABLED 71127800 0 - - -
pl user1_pl2 user1_vol ENABLED 71127800 - ACTIVE - -
sd user1_sd2 user1_pl2 ENABLED 71127800 0 - - -

Disk group: user2

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg user2 user2 - - - - - -

dm user201 dsk5 - 17769412 - - - -
dm user202 dsk11 - 17769412 - - - -

v user2_vol fsgen ENABLED 17769400 - ACTIVE - -
pl user2_pl1 user2_vol ENABLED 17769400 - ACTIVE - -
sd user2_sd1 user2_pl1 ENABLED 17769400 0 - - -
pl user2_pl2 user2_vol ENABLED 17769400 - ACTIVE - -
sd user2_sd2 user2_pl2 ENABLED 17769400 0 - - -

[Markus Schibli]

Hi Brendan

I don't know on which Tru64 version and patchlevel you're on, but in v5.1b there's a bug which looks quite similair to yours, and which was solved in pk5/bl26.

Here's the description of the fixed problem:
***
This patch fixes a problem whereby LSM mirrored volumes were incorrectly
left in the SYNC state after a cluster member failure and simultaneous
underlying disk storage failure.
LSM's 'volume' command is executed after a cluster member failure to recover
LSM mirrored volumes it was actively using.

Such volumes were put in the SYNC state before checking for the number of
usable plexes. If this count has been reduced to 1 because of simultaneous
underlying disk failures, the volume will not be resynchronized and the SYNC
state will be erroneously set.

# volprint -htA
Disk group: rootdg

DG NAME NCONFIG NLOG MINORS GROUP-ID
DM NAME DEVICE TYPE PRIVLEN PUBLEN STATE
V NAME USETYPE KSTATE STATE LENGTH READPOL PREFPLEX
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT NCOL/WID MODE
SD NAME PLEX DISK DISKOFFS LENGTH [COL/]OFF DEVICE MODE

dg rootdg default default 0 1103137167.1026.train

dm dsk7 dsk7 sliced 4096 17765065 -
dm dsk30 - - - - NODEVICE

v t1 fsgen ENABLED SYNC 2097152 SELECT -
pl t1-01 t1 ENABLED ACTIVE 2097152 CONCAT - RW
sd dsk7-01 t1-01 dsk7 0 2097152 0 dsk7 ENA
pl t1-02 t1 DISABLED NODEVICE 2097152 CONCAT - RW
sd dsk30-01 t1-02 dsk30 0 2097152 0 - RLOC
pl t1-03 t1 ENABLED ACTIVE LOGONLY CONCAT - RW
sd dsk7-02 t1-03 dsk7 2097152 65 LOG dsk7 ENA

With no resynchronization taking place, the volume is left in the SYNC state
even after the mirror has been restored. This means the volume will be left
in read-writeback mode which will affect read performance.

Without this fix, the volume would have to stopped and started again to
clear this state. If the volume cannot be stopped because a mounted filesystem
is using that storage, the cluster would have to be rebooted to clear the
state.
***

If you're not at pk5/bl26 please go on and install T64V51BB26AS0005-20050502

kind regards
Markus

[Brendan Murphy_5]

Hi Marcus,
Thanks for the info. The system is indeed T64 V5.1B TCR 5.1B & I believe it is running PK4. I'll check this & upgrade.

[Brendan Murphy_5]

Thanks Marcus... Will schedule PK5 install .. assuming the long awaited PK6 doesn't turn up;-)