From: Martin Wilck <mwilck@xxxxxxxx>
This reverts commit adf551f601596aac0774884c8a5b6b0f74721cfc.
Since 2b1fe13 ("libmultipath: allow force reload with no active paths"),
reloading maps with pending paths is no issue any more. The path checker
is going to set the correct status of the path quickly. If the paths are
down, I/O will eventually fail.
Sleeping with the vecs lock held is undesirable though, so avoid waiting
in setup_map().
Reviewed-by: Benjamin Marzinski <bmarzins@xxxxxxxxxx>
Signed-off-by: Martin Wilck <mwilck@xxxxxxxx>
---
libmultipath/configure.c | 63 +---------------------------------------
1 file changed, 1 insertion(+), 62 deletions(-)
diff --git a/libmultipath/configure.c b/libmultipath/configure.c
index 7f32da2..6745976 100644
--- a/libmultipath/configure.c
+++ b/libmultipath/configure.c
@@ -43,10 +43,6 @@
#include "sysfs.h"
#include "io_err_stat.h"
-/* Time in ms to wait for pending checkers in setup_map() */
-#define WAIT_CHECKERS_PENDING_MS 10
-#define WAIT_ALL_CHECKERS_PENDING_MS 90
-
/* group paths in pg by host adapter
*/
int group_by_host_adapter(struct pathgroup *pgp, vector adapters)
@@ -260,42 +256,11 @@ int rr_optimize_path_order(struct pathgroup *pgp)
return 0;
}
-static int wait_for_pending_paths(struct multipath *mpp,
- struct config *conf,
- int n_pending, int goal, int wait_ms)
-{
- static const struct timespec millisec =
- { .tv_sec = 0, .tv_nsec = 1000*1000 };
- int i, j;
- struct path *pp;
- struct pathgroup *pgp;
- struct timespec ts;
-
- do {
- vector_foreach_slot(mpp->pg, pgp, i) {
- vector_foreach_slot(pgp->paths, pp, j) {
- if (pp->state != PATH_PENDING)
- continue;
- pp->state = get_state(pp, conf,
- 0, PATH_PENDING);
- if (pp->state != PATH_PENDING &&
- --n_pending <= goal)
- return 0;
- }
- }
- ts = millisec;
- while (nanosleep(&ts, &ts) != 0 && errno == EINTR)
- /* nothing */;
- } while (--wait_ms > 0);
-
- return n_pending;
-}
-
int setup_map(struct multipath *mpp, char **params, struct vectors *vecs)
{
struct pathgroup * pgp;
struct config *conf;
- int i, n_paths, marginal_pathgroups;
+ int i, marginal_pathgroups;
char *save_attr;
/*
@@ -394,7 +359,6 @@ int setup_map(struct multipath *mpp, char **params, struct vectors *vecs)
if (marginal_path_check_enabled(mpp))
start_io_err_stat_thread(vecs);
- n_paths = VECTOR_SIZE(mpp->paths);
/*
* assign paths to path groups -- start with no groups and all paths
* in mpp->paths
@@ -409,31 +373,6 @@ int setup_map(struct multipath *mpp, char **params, struct vectors *vecs)
if (group_paths(mpp, marginal_pathgroups))
return 1;
- /*
- * If async state detection is used, see if pending state checks
- * have finished, to get nr_active right. We can't wait until the
- * checkers time out, as that may take 30s or more, and we are
- * holding the vecs lock.
- */
- if (conf->force_sync == 0 && n_paths > 0) {
- int n_pending = pathcount(mpp, PATH_PENDING);
-
- if (n_pending > 0)
- n_pending = wait_for_pending_paths(
- mpp, conf, n_pending, 0,
- WAIT_CHECKERS_PENDING_MS);
- /* ALL paths pending - wait some more, but be satisfied
- with only some paths finished */
- if (n_pending == n_paths)
- n_pending = wait_for_pending_paths(
- mpp, conf, n_pending,
- n_paths >= 4 ? 2 : 1,
- WAIT_ALL_CHECKERS_PENDING_MS);
- if (n_pending > 0)
- condlog(2, "%s: setting up map with %d/%d path checkers pending",
- mpp->alias, n_pending, n_paths);
- }
-
/*
* ponders each path group and determine highest prio pg
* to switch over (default to first)
--
2.33.1
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/dm-devel
