|
Hi Reinecke
Description of problem:
1.
As network issue, System is disconnected from the Storage.
2.
When network recover, we found two path are in “ghost” state
3.
We need to register PRKEY. But it failed when register PRKEY through the two “ghost” path.
4.
It seems wired, maybe the Storage hardware have issue. but now we want to work around this issue in multipath first.
5.
Our work around solution is as following
a)
Revert this patch “Return 'ghost' state when port is in standby”(5da642ff71254ecda685e82c681e78cab4c75b87)
b)
Then all PATH_GHOST will be treated as PATH_FAILED
c)
and multipath will only register PRKEY by PATH_UP pathes.
Questions:
so we want to consult whether there some potential issue after we revert the patch?
We know it will lead to some performance and reliability loss. We mainly worry about
whether there are some potential functional and fundamental issues?
Thanks a lot.
Environment Info: mpathal (2e0176ad6309077166c9ce90033bfa248_1) dm-2 TestSAN size=20G features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=50 status=active
|- 8:0:5:1 sddg 8:96 active ready running
`- 9:0:6:1 sdwh 12:112 active ready running
`-+- policy='round-robin 0' prio=10 status=enabled
|- 10:0:5:1 sdga 21:96 active ready running
`- 11:0:6:1 sdhb 81:112 active ready running
`-+- policy='round-robin 0' prio=1 status= enabled
|- 12:0:5:1 sdgj 15:6 active ghost running
`- 13:0:6:1 sdhk 16:112 active ghost running
defaults {
path_checker tur
no_path_retry 18
path_grouping_policy group_by_prio
prio const
deferred_remove yes
uid_attribute "ID_SERIAL"
reassign_maps no
failback immediate
log_checker_err once
}
devices {
device {
vendor "TestSAN"
product "LU"
path_grouping_policy group_by_prio
no_path_retry 30
prio alua
path_checker tur
path_selector "round-robin 0"
hardware_handler "0"
failback 15
}
} |
-- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
