The way that multipathd was handling priority refreshes had some issues.
Some of the multipath prioritizers can hang when run on a failed path.
Multipathd was only skipping paths in PATH_DOWN, but there are other
states where the prioritizer is also likely to hang, such as
PATH_TIMEOUT, PATH_SHAKY, and to a lesser extent, PATH_DELAYED.
Also, before the recent patch splitting the priority updating from the
path checking, multipathd wasn't consistent with which states would
cause
paths to get their priorities updated. If a path changed its state to
anything other than PATH_UP or PATH_GHOST, it wouldn't get its priority
updated. But if a path kept the same state its priority would get
updated as long at the state wasn't PATH_DOWN.
For safety's sake, a path's priority should only get refreshed when its
in the PATH_UP or PATH_GHOST state. This shouldn't cause problems. Only
paths that are in the PATH_UP or PATH_GHOST state are usable by the
kenel and contibute to the pathgroup's priority.
Signed-off-by: Benjamin Marzinski <bmarz...@redhat.com>
---
multipathd/main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/multipathd/main.c b/multipathd/main.c
index 75bc0620..91806e9d 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -2162,7 +2162,7 @@ static bool update_prio(struct multipath *mpp, bool
refresh_all)
vector_foreach_slot (mpp->pg, pgp, i) {
vector_foreach_slot (pgp->paths, pp, j) {
- if (pp->state == PATH_DOWN)
+ if (pp->state != PATH_UP && pp->state != PATH_GHOST)
continue;
if (!refresh_all &&
pp->is_checked != CHECK_PATH_CHECKED) {
@@ -2186,7 +2186,7 @@ static bool update_prio(struct multipath *mpp, bool
refresh_all)
*/
vector_foreach_slot (mpp->pg, pgp, i) {
vector_foreach_slot (pgp->paths, pp, j) {
- if (pp->state == PATH_DOWN)
+ if (pp->state != PATH_UP && pp->state != PATH_GHOST)
continue;
if (pp->is_checked == CHECK_PATH_CHECKED)
continue;
--
2.45.0
