Hannes,
I merged the patchset.
Can you post a patch to set return values to the multipathd/main.c::patch_check() exit points where it is not present.
This 11/12 patch adds one, and the "proactive path removal" patch adds the other.
This requirement was introduced by your earlier "check timing" patch.
Thanks,
Christophe Varoqui
On Fri, Dec 13, 2013 at 1:14 PM, Hannes Reinecke <hare@xxxxxxx> wrote:
When a path is discovered in 'blocked' state it won't be added
to the path list as we cannot get the path uuid.
And any further check will fail here as the uuid won't be
recovered. So add a check in checkerloop() to re-add the
path if it gets out of a 'blocked' state.
Signed-off-by: Hannes Reinecke <hare@xxxxxxx>
---
multipathd/main.c | 14 +++++++++++---
1 file changed, 11 insertions(+), 3 deletions(-)
diff --git a/multipathd/main.c b/multipathd/main.c
index 7058147..cc5cf67 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -1111,9 +1111,6 @@ check_path (struct vectors * vecs, struct path * pp)
int chkr_new_path_up = 0;
int oldchkrstate = pp->chkrstate;
- if (!pp->mpp)
- return;
-
if (pp->tick && --pp->tick)
return; /* don't check this path yet */
@@ -1139,6 +1136,17 @@ check_path (struct vectors * vecs, struct path * pp)
pathinfo(pp, conf->hwtable, 0);
return;
}
+ if (!pp->mpp) {
+ if (!strlen(pp->wwid) &&
+ (newstate == PATH_UP || newstate == PATH_GHOST)) {
+ condlog(2, "%s: add missing path", pp->dev);
+ if (pathinfo(pp, conf->hwtable, DI_ALL) == 0) {
+ ev_add_path(pp, vecs);
+ pp->tick = 1;
+ }
+ }
+ return;
+ }
/*
* Async IO in flight. Keep the previous path state
* and reschedule as soon as possible
--
1.8.1.4
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
-- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
