multipath: fix setting of fast_io_fail_tmo
authorPetr Uzel <petr.uzel@suse.cz>
Tue, 16 Jul 2013 07:13:19 +0000 (09:13 +0200)
committerChristophe Varoqui <christophe.varoqui@opensvc.com>
Tue, 16 Jul 2013 19:56:58 +0000 (21:56 +0200)
Setting fast_io_fail_tmo to the same value as dev_loss_tmo is
not allowed by the kernel. Increase dev_loss_tmo by 1 in such
cases to make it strictly greated than fast_io_fail_tmo.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
libmultipath/discovery.c

index d31ec46..7035a43 100644 (file)
@@ -366,7 +366,7 @@ sysfs_set_rport_tmo(struct multipath *mpp, struct path *pp)
                        goto out;
                }
                if (mpp->fast_io_fail >= tmo) {
-                       snprintf(value, 16, "%u", mpp->fast_io_fail);
+                       snprintf(value, 16, "%u", mpp->fast_io_fail + 1);
                }
        } else if (mpp->dev_loss > 600) {
                condlog(3, "%s: limiting dev_loss_tmo to 600, since "