ASM Disk Group Problem

[Chucks_k]

Hi all,

We are running 10.2.03 on Redhat 3.

We have ASM installed and are using EMC San for storage. We had a hardware panic and 3 (these were part of the failover group) of the 6 disks available went offline but reappeared as blank group name:

PATH                           NAME                           STATE
------------------------------ ------------------------------ --------
/dev/raw/raw6                                                 NORMAL
/dev/raw/raw5                                                 NORMAL
/dev/raw/raw4                                                 NORMAL
                               DGGROUP_0005                   FORCING
                               DGGROUP_0004                   FORCING
                               DGGROUP_0003                   FORCING
/dev/raw/raw3                  DGGROUP_0002                   NORMAL
/dev/raw/raw2                  DGGROUP_0001                   NORMAL
/dev/raw/raw1                  DGGROUP_0000                   NORMAL

As can from above, raw6,5,4 now appear with no name but are online, the blanks are all offline. I did a alter diskgroup check all and this did not throw up any errors. Now how do i get the raw6,5,4 to be part of a failover group?

SQL> select failgroup, name from v$asm_disk;

FAILGROUP                      NAME
------------------------------ ------------------------------



                               DGGROUP_0005
                               DGGROUP_0004
                               DGGROUP_0003
PRIMARY                        DGGROUP_0002
PRIMARY                        DGGROUP_0001
PRIMARY                        DGGROUP_0000

Also does this mean the even though these disks are not part of the failure group, normal redundancy is still availabe as all 6 are still online? Any help will be greatly appreciated

Thanks in advance
Chucks

[BV1963]

Have your sa to check if they came back with the same device id

[Chucks_k]

Hi,

Thanks for that. My SA did mention that the disks have 'multiple' paths, not sure if thats of any relevance. What happens if the device id is different?

Thank you in advance

[BV1963]

I am not an SA, but I think it may not be recognized as a same device.

[Chucks_k]

Right, but in terms of Oracle ASM does this mean that we dont have redundacy. It looks as if 6 disks are online as per the output above which leads me to believe the redundancy is there albiet in an unconvential manner?

[pando]

you should have used asmlib because you get device persistent name with it and not with raw devices

raw4,5 and 6 are probably not the ASM disks which you think they are

show us the output of device names

[Chucks_k]

Hi Pando,

Thanks for that, I got not get ASMLib to work. ASM could not discover the disks available for some reason, hence that is why i went down the raw device route. But the ouput of device names do you mean config of the /etc/sysconfig/rawdevices file?

In /etc/sysconfig/rawdevices, I bound the devices as follows:

/dev/raw/raw1 /dev/emcpowerd1
/dev/raw/raw2 /dev/emcpowerh1
/dev/raw/raw3 /dev/emcpowerg1
/dev/raw/raw4 /dev/emcpowerc1
/dev/raw/raw5 /dev/emcpowere1
/dev/raw/raw6 /dev/emcpowerf1

Thanks in advance

[pando]

show us the output of

select HEADER_STATUS,MODE_STATUS,STATE,LABEL from v$asm_disk;


you could not configure asmlib probably because you need to modify /etc/sysconfig/oracleasm and asmlib v2.0

[Chucks_k]

Hi,

This is the output u requested:

SQL> select HEADER_STATUS,MODE_STATUS,STATE,LABEL from v$asm_disk;

HEADER_STATU MODE_ST STATE    LABEL
------------ ------- -------- -------------------------------
MEMBER       ONLINE  NORMAL
MEMBER       ONLINE  NORMAL
MEMBER       ONLINE  NORMAL
CANDIDATE    OFFLINE FORCING
CANDIDATE    OFFLINE FORCING
CANDIDATE    OFFLINE FORCING
MEMBER       ONLINE  NORMAL
MEMBER       ONLINE  NORMAL
MEMBER       ONLINE  NORMAL

9 rows selected

Thanks,
Chucks

[pando]

check ASM alert and see if it shows any errors mounting the disks

another way is use kfed and look the disk headers which shows ASM metadata such as group name and disk name

to build kfed run

cd $ORACLE_HOME/lib
make -f ins_rdbms.mk ikfed

kfed read /dev/raw/raw4