[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
| Title: | ase |
|
| Moderator: | SMURF::GROSSO |
|
| Created: | Thu Jul 29 1993 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2114 |
| Total number of notes: | 7347 |
1895.0. " LSM disks online from 2 members " by EVTAI1::POUSSARD () Tue Feb 25 1997 06:37
Hi,
Did someone experiment this following problem with a group disk on a
shared bus online to 2 members instead of only one ??
My configuration is : tri host UNIX ASE cluster ( ASE 1.4, DU 4.0B )
The customer complains that all disks from a disk group belonging to
an ASE service, are online from 2 members. ( voldisk list )
He asks me for a scenario which could give this result ( we assume
that customer didn't execute LSM specific commands like voldg import or
voldisk online commands )
Here is the trace of the problem
The 3 machines call prod1, prod2, prod3
service 'exp' is running on prod3 . So on prod3, we see correctly
the disks through LSM
The disks belonging to 'exp' are still online on prod2 , which is
abnormal. See below.
Obtaining ASE Status
m) Display the status of the members
s) Display the status of a service
l) Display the location of logger(s)
v) Display the level of logging
x) Exit to the Main Menu ?) Help
Enter your choice [x]: s
Service Status
Select the service whose status you want to display:
1) exp on prod3
2) qpsvar on prod3
3) qps on prod2
x) Exit to previous menu ?) Help
Enter your choice [x]: 1
Status for DISK service `exp`
Status: Relocate: Placement Policy: Favored Member(s):
on prod3 no Favor Member(s) prod3,prod2,prod1
Storage configuration for DISK service `exp`
Mount Table (device, mount point, type, options)
EXP#gestion /gestion advfs rw,groupquota,userquota
EXP#exploita /exploita advfs rw,groupquota,userquota
Advfs Configuration
Domain: Volume(s):
EXP /dev/vol/expdg/exp100-vol /dev/vol/expdg/exp630-vol
LSM Configuration
Disk Group: Device(s):
expdg rz27 rz19 rzh25 rzh17
Press 'Return' to continue:
Service Status
Select the service whose status you want to display:
So,(rz27 rz19 rzh25 rzh17) belong to service 'exp' which run on machine
'prod3'. So on prod3, voldisk and volprint gives coherent results
prod3 > voldisk list
# voldisk list
DEVICE TYPE DISK GROUP STATUS
re0h simple re0h rootdg online
rz19 sliced exp100ar expdg online
rz20 sliced - - offline
rz27 sliced exp100 expdg online
rz28 sliced - - offline
rzb19 sliced - - offline
rzb20 sliced - - offline
rzb27 sliced - - offline
rzb28 sliced - - offline
rzc19 sliced - - offline
rzc20 sliced - - offline
rzc27 sliced - - offline
rzc28 sliced - - offline
rzd18 sliced var620ar vardg online
rzd20 sliced var110ar vardg online
rzd26 sliced var620 vardg online
rzd28 sliced var110 vardg online
rze18 sliced - - offline
rze26 sliced - - offline
rzf18 sliced - - offline
rzf26 sliced - - offline
rzg17 sliced - - offline
rzg18 sliced - - offline
rzg25 sliced - - offline
rzg26 sliced - - offline
rzh17 sliced exp630ar expdg online
rzh18 sliced var430ar vardg online
rzh25 sliced exp630 expdg online
rzh26 sliced var430 vardg online
prod3 > volprint -htA
# volprint -htA
Disk group: rootdg
DG NAME GROUP-ID
DM NAME DEVICE TYPE PRIVLEN PUBLEN PUBPATH
V NAME USETYPE KSTATE STATE LENGTH READPOL PREFPLEX
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT ST-WIDTH
MODE
SD NAME PLEX PLOFFS DISKOFFS LENGTH DISK-NAME DEVICE
dg rootdg 855268713.1025.prod3
dm re0h re0h simple 1024 10869 /dev/rre0h
Disk group: expdg
DG NAME GROUP-ID
DM NAME DEVICE TYPE PRIVLEN PUBLEN PUBPATH
V NAME USETYPE KSTATE STATE LENGTH READPOL PREFPLEX
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT ST-WIDTH
MODE
SD NAME PLEX PLOFFS DISKOFFS LENGTH DISK-NAME DEVICE
dg expdg 789238287.1798.prod2
dm exp100 rz27 sliced 512 8377500 /dev/rrz27h
dm exp100ar rz19 sliced 512 8377500 /dev/rrz19h
dm exp630 rzh25 sliced 512 8377500 /dev/rrzh25h
dm exp630ar rzh17 sliced 512 8377500 /dev/rrzh17h
v exp100-vol fsgen ENABLED ACTIVE 8377499 SELECT -
pl exp100-pl exp100-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd exp100-sd exp100-pl 0 0 8377499 exp100 rz27
pl exp100ar-pl exp100-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd exp100ar-sd exp100ar-pl 0 0 8377499 exp100ar rz19
v exp630-vol fsgen ENABLED ACTIVE 8377499 SELECT -
pl exp630-pl exp630-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd exp630-sd exp630-pl 0 0 8377499 exp630 rzh25
pl exp630ar-pl exp630-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd exp630ar-sd exp630ar-pl 0 0 8377499 exp630ar rzh17
Disk group: vardg
DG NAME GROUP-ID
DM NAME DEVICE TYPE PRIVLEN PUBLEN PUBPATH
V NAME USETYPE KSTATE STATE LENGTH READPOL PREFPLEX
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT ST-WIDTH
MODE
SD NAME PLEX PLOFFS DISKOFFS LENGTH DISK-NAME DEVICE
dg vardg 789238035.1780.prod2
dm var110 rzd28 sliced 512 8377500 /dev/rrzd28h
dm var110ar rzd20 sliced 512 8377500 /dev/rrzd20h
dm var430 rzh26 sliced 512 8377500 /dev/rrzh26h
dm var430ar rzh18 sliced 512 8377500 /dev/rrzh18h
dm var620 rzd26 sliced 512 8377500 /dev/rrzd26h
dm var620ar rzd18 sliced 512 8377500 /dev/rrzd18h
v var110-vol fsgen ENABLED ACTIVE 8377499 SELECT -
pl var110-pl var110-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var110-sd var110-pl 0 0 8377499 var110 rzd28
pl var110ar-pl var110-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var110ar-sd var110ar-pl 0 0 8377499 var110ar rzd20
v var430-vol fsgen ENABLED ACTIVE 8377499 SELECT -
pl var430-pl var430-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var430-sd var430-pl 0 0 8377499 var430 rzh26
pl var430ar-pl var430-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var430ar-sd var430ar-pl 0 0 8377499 var430ar rzh18
v var620-vol fsgen ENABLED ACTIVE 8377499 SELECT -
pl var620-pl var620-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var620-sd var620-pl 0 0 8377499 var620 rzd26
pl var620ar-pl var620-vol ENABLED ACTIVE 8377499 CONCAT - RW
sd var620ar-sd var620ar-pl 0 0 8377499 var620ar rzd18
On prod2, we see ( rz27,rz19,rzh25,rzh17 ) also online which is abnormal
prod2 > voldisk list
DEVICE TYPE DISK GROUP STATUS
re0h simple re0h rootdg online
rz19 sliced - - online
rz20 sliced pub400ar pubdg online
rz27 sliced - - online
rz28 sliced pub400 pubdg online
rzb19 sliced qps200ar qpsdg online
rzb20 sliced pub500ar pubdg online
rzb27 sliced qps200 qpsdg online
rzb28 sliced pub500 pubdg online
rzc19 sliced qps300ar qpsdg online
rzc20 sliced pub600ar pubdg online
rzc27 sliced qps300 qpsdg online
rzc28 sliced pub600 pubdg online
rzd18 sliced - - error
rzd20 sliced - - error
rzd26 sliced - - error
rzd28 sliced - - error
rze18 sliced pub130ar pubdg online
rze26 sliced pub130 pubdg online
rzf18 sliced pub230ar pubdg online
rzf26 sliced pub230 pubdg online
rzg17 sliced qps530ar qpsdg online
rzg18 sliced pub330ar pubdg online
rzg25 sliced qps530 qpsdg online
rzg26 sliced pub330 pubdg online
rzh17 sliced - - online
rzh18 sliced - - error
rzh25 sliced - - online
rzh26 sliced - - error
So, I don't know how the customer reached this situation ans asks me for a
possible ASE scenario.
I can't say what is happening. The only way I said him is to reboot
proerly the cluster and do some tests to see if we can reproduce this
situation.
Thanks for your help
Cross posted in DIGITAL UNIX and ASE conference
Gilles
| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1895.1 | Online/Offline is only a FLAG... | BACHUS::DEVOS | Manu Devos DEC/SI Brussels 856-7539 | Thu Feb 27 1997 07:33 | 21 |
| Gilles,
The situation you describe is absolutely not dangerous. The fact theat
the disks are online is not implying they are seen from the system. In
LSM, the online/offline flag is simply a software flag which
allow/disallow to use the disks. This falg is stored in the rootdg
diskgroup for each DiskAccess names (DA). You can reach this situation
for example when exp is running on a system and this system is
crashing. Thus, ASE had no chance to place the LSM disk offline and
another member has taken the service over.
I even think that rebooting the system will not place them offline. You
should type the following command:
# voldisk offline rz17 rz25 rz19 rzh27
Don't worry, be happy !
Manu.
|
| 1895.2 | I found the problem | EVTAI1::POUSSARD | | Mon Mar 03 1997 07:36 | 12 |
| Finally, I found the reason of my problem
If putting a service offline without success and still asking to
ASE to put the service offline, ASE will put the service offline, but
the LSM disk group are still importted. You musrt then manually deport
the disk group and put the disk offline like ASE does when it put
successfully offline. Doing not this, and putting the service online,
and this service going to another preferred memeber allow to reach this
situation
Gilles.
|