topic Re: reflecting new lun size from storage w/o rebooting in Operating System - Linux

reflecting new lun size from storage w/o rebooting

glennn — Thu, 10 Sep 2009 04:15:17 GMT

hi.

has anyone here experienced doing this:

you have a lun connected thru FC to an eva or msa or any storage device for that matter, and you increased the lun in the storage device, then you're able to reflect the new lun size w/o rebooting? i've tried tools like hp_rescan and qlogic's rescan_scsi_bus.sh tool w/c doesn't help. what i do is to check the scsi path, for example 0:0:0:1, then do this: echo "remove-single-device 0 0 0 1" > /proc/scsi/scsi && echo "add-single-device 0 0 0 1" > /proc/scsi/scsi. It does work. But naturally, the /dev/sd* mapped to it will change. I use the multipath mapped /dev/dm* device w/ lvm2, so i'm able to get away with the changing /dev/sd.

But if any of you guys have a better way of doing it, i'd be happy to check it out. Mine is an SLES9 SP4. I've checked out other forums, and most say that you really need a reboot for sles9 and even 10.

thanks and best regards.

Re: reflecting new lun size from storage w/o rebooting

Michael Leu — Fri, 11 Sep 2009 06:43:57 GMT

For new LUNs in SLES10 we do a
echo "- - -" > /sys/class/scsi_host/host0/scan
echo "- - -" > /sys/class/scsi_host/host1/scan
as described here:
http://www.novell.com/documentation/sles10/stor_evms/?page=/documentation/sles10/stor_evms/data/scandev.html

I have never tried to grow a LUN, just added new ones.. not sure if this helps.

Re: reflecting new lun size from storage w/o rebooting

Ivan Ferreira — Fri, 11 Sep 2009 10:23:35 GMT

As Michael said, normally with the use of LVM you will add new LUNs to existing volume groups instead of increasing the LUN size. Have you also tried the partprobe command?

Re: reflecting new lun size from storage w/o rebooting

skt_skt — Sat, 12 Sep 2009 15:16:14 GMT

is there a specific reason to strict to grow the existing LUN? I have seen some doc with EMC and LINUX combination talking describing how to do this online.