| Message ID | 1512672546-12785-12-git-send-email-bmarzins@redhat.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Delegated to: | christophe varoqui |
| Headers | show |
On Thu, 2017-12-07 at 12:49 -0600, Benjamin Marzinski wrote: > If a path's checker state is down before and after a path check, but > the > path's dmstate is active, mutipath won't update the dmstate. It only > updates the dmstate when the path first fails. This can cause the > kernel to try known faulty paths, if the multipath device was > reloaded > outside of multipathd. check_path() already checks for and deals > with a > similar case where the path's checker state is up before and after a > path check, but the dmstate is failed. It should do the same thing > for > faulty paths. > > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> > --- Reviewed-by: Martin Wilck <mwilck@suse.com>
diff --git a/multipathd/main.c b/multipathd/main.c index a420c83..16e4fdf 100644 --- a/multipathd/main.c +++ b/multipathd/main.c @@ -1770,16 +1770,21 @@ check_path (struct vectors * vecs, struct path * pp, int ticks) pp->tick = pp->checkint; } } - else if (newstate == PATH_DOWN) { - int log_checker_err; + else if (newstate != PATH_UP && newstate != PATH_GHOST) { + if (pp->dmstate == PSTATE_ACTIVE || + pp->dmstate == PSTATE_UNDEF) + fail_path(pp, 0); + if (newstate == PATH_DOWN) { + int log_checker_err; - conf = get_multipath_config(); - log_checker_err = conf->log_checker_err; - put_multipath_config(conf); - if (log_checker_err == LOG_CHKR_ERR_ONCE) - LOG_MSG(3, checker_message(&pp->checker)); - else - LOG_MSG(2, checker_message(&pp->checker)); + conf = get_multipath_config(); + log_checker_err = conf->log_checker_err; + put_multipath_config(conf); + if (log_checker_err == LOG_CHKR_ERR_ONCE) + LOG_MSG(3, checker_message(&pp->checker)); + else + LOG_MSG(2, checker_message(&pp->checker)); + } } pp->state = newstate;
If a path's checker state is down before and after a path check, but the path's dmstate is active, mutipath won't update the dmstate. It only updates the dmstate when the path first fails. This can cause the kernel to try known faulty paths, if the multipath device was reloaded outside of multipathd. check_path() already checks for and deals with a similar case where the path's checker state is up before and after a path check, but the dmstate is failed. It should do the same thing for faulty paths. Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com> --- multipathd/main.c | 23 ++++++++++++++--------- 1 file changed, 14 insertions(+), 9 deletions(-)