When removing a LUN from a Red Hat Enterprise Linux 5 (RHEL 5) server, I ran into this interesting issue:
[root@server ~]# multipath -ll
sdat: checker msg is "tur checker reports path is down"
sdbt: checker msg is "tur checker reports path is down"
sdct: checker msg is "tur checker reports path is down"
sdt: checker msg is "tur checker reports path is down"
This is annoying and problematic since running
pvs or something similar will cause the system to hang as it tries to access the missing LUN. It's briefly covered below, but see also the official documentation:
http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Online_Storage_Reconfiguration_Guide/removing_devices.html
- Stop using the volume and umount
- Remove it from software RAID or LVM volume group
- Use multipath -l to get the block devices names (e.g., sda, sdb, etc.)
- Use multipath -f mpathX
- To flush outstanding I/O ("sync"), use blockdev –flushbufs
device
- Make sure applications and things are no longer referencing the device(s)
- echo 1 > /sys/block/
device-name/device/delete
I had to clean up lots of luns, and was lazy..
ReplyDeleteWARNING: Potentially dangerous!!
#!/bin/bash
# script to remove luns from device-mapper and scsi layer
#
if [ -z $1 ]
then
echo "usage: ./removelun "
echo "example: ./removelun mpath42"
exit
fi
MPATH=$1
echo query device $MPATH..
DEVICES=`multipath -ll $MPATH | grep ':.:.:' | awk '{print $3}' `
echo $DEVICES
echo flushing multipath device $MPATH
multipath -f $MPATH
echo flush outstanding i/o to sd devices ..
for DEVICE in $DEVICES
do
echo $DEVICE..
blockdev --flushbufs /dev/$DEVICE
done
echo deleting sd devices..
for DEVICE in $DEVICES
do
echo $DEVICE..
echo 1 > /sys/block/$DEVICE/device/delete
done
echo done, now you can unpresent on disk array