Sporadic ASMLIB error function on RedHat 2.6.9-42.ELsmp
564759Mar 1 2007 — edited Jul 6 2007RedHat ES4 2.6.9-42.ELsmp
(2) Qlogic 2340 hba
Xiotech SAN
10.2.0.3
oracleasm-2.6.9-42.ELsmp-2.0.3-1
oracleasm-support-2.0.3-1
oracleasmlib-2.0.2-1
/etc/init.d/oracleasm commands all work fine, never errors.
But when the dba does things like this:
ALTER DATABASE DATAFILE '+ASM_DATA1/fed1kc/datafile/undotbs2.264.615649287' AUTOEXTEND ON NEXT 102400K MAXSIZE 2048000K
or
CREATE UNDO TABLESPACE "UNDOTBS2" DATAFILE SIZE 25600K AUTOEXTEND ON MAXSIZE UNLIMITED EXTENT MANAGEMENT LOCAL
oracle trace logs this:
ORA-15186: ASMLIB error function = [asm_io], error = [3], mesg = [I/O Error]
ORA-19502: write error on file "+ASM_DATA1/fed1kc/datafile/undotbs2.264.615649287", blockno 5120 (blocksize=8192)
ORA-15186: ASMLIB error function = [], error = [], mesg = []
-I get this in /var/log/messages at the same second:
kernel: end_request: I/O error, dev sdb, sector 3505215
-sdb is the ASM partition
If dba tries the same command 3 or 4 times, it finally goes through and completes.
i am thinking maybe set the hba card speed manually instead of auto-sense, or some things like that.
anybody got ideas?
thanks