Hi there, using HP-UX 11 in a L2000 SG-Cluster i've got a problem with arraycfg, look at that prompt, please:
hera:/# arraycfg -L 0 -d -V $ID DeleteLUN(0) command received DeleteLUN(0) command started DeleteLUN(0) command complete Error in command execution, "RMT_ARMERRORSTATUS_MSG"
ARMERR : ERR_COMMAND_FAILED ARMERR Qual : DELETE_LUN MODULE_CODE_ID : SUBSYSTEM OS LAYER ID : HP-UX COMMAND STATE : An OEM errno was returned ERROR NUMBER : 16 OEM MESSAGE : "Device busy"
arraycfg: Error in command execution
first i shut down the SG-package, then i did a vgexport to delete the VG in that LUN. The i wanted to delete the LUN using the arraycfg command. This procedure worked with the other LUNs fine. What does this error mean ?
After opening a HP-Call i know what the device made busy: The vg in that LUN was the FIRST_CLUSTER_LOCK_VG ! Halting the cluster fixed that problem ...
I had created a bunch of luns and when I got them into vgdisplay I saw the PE count was 5998, I???d rather see 6000, a nice even number. I tried to delete the luns and format the array, reset and shutdown the array, it was hung up on delete or format commands. I was getting the same above errors.
Solution: I didn???t have cluster locks, but I did try to skip doing all the vgreduce commands, so I had simply removed the /dev/vg01/group and moved the /etc/lvmtab (remember always move that file never delete it, you don???t know if you???ll need it in the next steps.) I even tried to vgscan -v to recreate lvmtab. Still no good, got the SCSI ???device busy???. So I mv???d the old lvmtab back in place and ran all the vgreduce commands, it failed on I/O errors on the last lun removed (I went in reverse order that I had added them.)
After the vg was reduced correctly I redid the lvmtab steps, saving it to another name. Ran the vgscan, suddenly my arraycfg commands worked again. -db5/02
Hi Jay, thanks for your assistance. I'm not sure if i understand your response right, but my problem had been solved by the Call Center in February. Regards Lothar
