| Message ID | 1248984835.18243.5.camel@chandra-ubuntu (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Delegated to: | christophe varoqui |
| Headers | show |
Chandra Seetharaman wrote: > Hi Christophe, > > I submitted this patch on Jul 2 > (http://marc.info/?l=dm-devel&m=124658334721911&w=2). Resending it. > > Only change is a field name from up_paths to enabled_paths. > > Hi Hannes, > > Need an ACK from you :-). > > regards, > > chandra > > ----------------------------------------------------------------------- > Failback happens only when the sum of priorities of all paths > (on the higher priority path group) is greater than the sum > of priorities of all paths on the lower priority path group. > > This leads into problems when there are more than one paths > in each of the path groups, and the sum of all paths in the > lower priority path group is greater than that of path priority > of a single high priority path. > > This patch fixes the problem by using average priority of > the path group to decide on which path group to switch over. > > Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Acked-by: Hannes Reinecke <hare@suse.de> Thanks for this, Chandra. Cheers, Hannes
Index: multipath-tools-mainline/libmultipath/structs.h =================================================================== --- multipath-tools-mainline.orig/libmultipath/structs.h +++ multipath-tools-mainline/libmultipath/structs.h @@ -202,6 +202,7 @@ struct pathgroup { long id; int status; int priority; + int enabled_paths; vector paths; char * selector; }; Index: multipath-tools-mainline/libmultipath/switchgroup.c =================================================================== --- multipath-tools-mainline.orig/libmultipath/switchgroup.c +++ multipath-tools-mainline/libmultipath/switchgroup.c @@ -14,13 +14,16 @@ path_group_prio_update (struct pathgroup int priority = 0; struct path * pp; + pgp->enabled_paths = 0; if (!pgp->paths) { pgp->priority = 0; return; } vector_foreach_slot (pgp->paths, pp, i) { - if (pp->state != PATH_DOWN) + if (pp->state != PATH_DOWN) { priority += pp->priority; + pgp->enabled_paths++; + } } pgp->priority = priority; } @@ -29,8 +32,9 @@ extern int select_path_group (struct multipath * mpp) { int i; - int highest = 0; + int max_priority = 0, avg_priority; int bestpg = 1; + int max_enabled_paths = 1; struct pathgroup * pgp; if (!mpp->pg) @@ -41,9 +45,18 @@ select_path_group (struct multipath * mp continue; path_group_prio_update(pgp); - if (pgp->priority > highest) { - highest = pgp->priority; - bestpg = i + 1; + if (pgp->enabled_paths) { + avg_priority = pgp->priority / pgp->enabled_paths; + if (avg_priority > max_priority) { + max_priority = avg_priority; + max_enabled_paths = pgp->enabled_paths; + bestpg = i + 1; + } else if (avg_priority == max_priority) { + if (pgp->enabled_paths > max_enabled_paths) { + max_enabled_paths = pgp->enabled_paths; + bestpg = i + 1; + } + } } } return bestpg;
Hi Christophe, I submitted this patch on Jul 2 (http://marc.info/?l=dm-devel&m=124658334721911&w=2). Resending it. Only change is a field name from up_paths to enabled_paths. Hi Hannes, Need an ACK from you :-). regards, chandra ----------------------------------------------------------------------- Failback happens only when the sum of priorities of all paths (on the higher priority path group) is greater than the sum of priorities of all paths on the lower priority path group. This leads into problems when there are more than one paths in each of the path groups, and the sum of all paths in the lower priority path group is greater than that of path priority of a single high priority path. This patch fixes the problem by using average priority of the path group to decide on which path group to switch over. Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> --- libmultipath/structs.h | 1 + libmultipath/switchgroup.c | 23 ++++++++++++++++++----- 2 files changed, 19 insertions(+), 5 deletions(-) -- dm-devel mailing list dm-devel@redhat.com https://www.redhat.com/mailman/listinfo/dm-devel