现象:
The server has incorrect device mappings.
The /etc/multipath.conf shows following wwid/alias pairs:
multipaths {
(snip)
multipath {
wwid <WWID1>
alias <ALIAS1>
}
multipath {
wwid <WWID2>
alias <ALIAS2>
}
(snip)
}
But "multipath-ll" shows
<ALIAS1> (<WWID2>) dm-22 NETAPP,LUN C-Mode >>> should be <ALIAS2> per setting in multipath.conf
<ALIAS2> (<WWID1>) dm-21 NETAPP,LUN C-Mode >>> should be <ALIAS1> per setting in multipath.conf
原因:
The logs shows multipathd already found the wrong mappings, but with no clue to what happened exactly.
Feb 24 12:07:28 hostname multipathd: <WWID1>: unable to rename <ALIAS2> to <ALIAS1> (<ALIAS1> is used by <WWID2>)
Feb 24 12:07:28 hostname multipathd: <WWID2>: unable to rename <ALIAS1> to <ALIAS2> (<ALIAS2> is used by <WWID1>)
处理方法:
You don't need full outage window, flush the specific mapping and then re-run multipath to trigger re-mapping.
In the example below, we can see the mapping is corrected.
# multipath -f <ALIAS1>
# multipath -f <ALIAS2>
# multipath -ll | egrep "dm-21|dm-22"
<ALIAS1> (<WWID1>) dm-21 NETAPP,LUN C-Mode
<ALIAS2> (<WWID2>) dm-22 NETAPP,LUN C-Mode
|