Solaris

Here is the information requested:

0> sccli
sccli: /dev/es/ses1: device reset detected
sccli: /dev/es/ses2: device reset detected
sccli: selected device /dev/rdsk/c2t0d0s2 [SUN StorEdge 3320 SN#0B37F3]
sccli>
sccli> show lun
Ch Tgt LUN ld/lv ID-Partition Assigned Filter Map
---------------------------------------------------------------------
1 0 0 ld0 5CE01D7A-00 Primary
1 0 1 ld0 5CE01D7A-01 Primary
1 0 2 ld0 5CE01D7A-02 Primary
3 0 0 ld0 5CE01D7A-00 Primary
3 0 1 ld0 5CE01D7A-01 Primary
3 0 2 ld0 5CE01D7A-02 Primary
sccli> show channel
Ch Type Media Speed Width PID / SID
--------------------------------------------
0 Drive SCSI 160M Wide 6 / 7
1 Host SCSI 160M Wide 0,2,3 / N/A
2 Drive SCSI ASYNC Narrow 6 / 7
3 Host SCSI 160M Wide 0 / N/A
4 RCCOM FC(L) N/A Narrow N/A / N/A
5 RCCOM FC(L) ASYNC Narrow N/A / N/A
6 Host LAN N/A Serial N/A / N/A
sccli> show logical-drive
LD LD-ID Size Assigned Type Disks Spare Failed Status
------------------------------------------------------------------------
ld0 5CE01D7A 204.35GB Primary RAID5 4 1 0 Good
Write-Policy Default StripeSize 128KB
sccli> show partition
LD/LV ID-Partition Size
-------------------------------------
ld0-00 5CE01D7A-00 1000MB
ld0-01 5CE01D7A-01 101.69GB
ld0-02 5CE01D7A-02 101.69GB
sccli>

Thanks for the help in advance

Can you log a case with Oracle instead? Cos your setup looks very similar to mine. And are both controllers of the same firmware level? This is important. If everything is ok, I don't see why Oracle should/can force a system reboot , which is abnormal and not right

Do you have your disk array simultaneously connected to two Sun hosts in order to act as database storage for 2-node Oracle 10g RAC configuration?
We do not have Oracle support at the moment! So I am trying to see if someone else has experience in this to match our configuration with. Thanks!

Yes my config is as such. You need to check whats the oracle is doing. If not, at least provide the log messages that you get (what the messages file captured before it rebooted).
What i can say is, its nothing with the OS. Its with the RAC (real app clusters) for Oracle thats causing this. You NEED Oracle support, seriously

The messages file did not show much details about the reboot; that particular period of the time is missed in the messages file!! It looks like a power outage (which we know is not), or the message queue is too long at the time it was shutown before the reboot can be logged. However, the ocssd.log file shows the following which gives an indication that the reboot was initiated by Oracle:

[ CSSD]2009-03-08 03:18:17.454 [16] >TRACE: clssnmPollingThread: Eviction started for node srt-hlr-b (2), flags 0x000d, state 3, wt4c 0
[ CSSD]2009-03-08 03:18:17.454 [18] >TRACE: clssnmDoSyncUpdate: Initiating sync 3
[ CSSD]2009-03-08 03:18:17.454 [18] >TRACE: clssnmSetupAckWait: Ack message type (11)

Thanks for your comments

Is it true that when Oracle "evict" a node, it forces it to reboot? When this happens, I have noticed that Solaris (messages log) does not capture anything related to this reboot as if it was a power outage!!

StarSol can be right too :/ it could happen the same during a power outage