diff options
| author | Kevin Vigor <kvigor@fb.com> | 2017-01-04 16:29:23 -0800 |
|---|---|---|
| committer | Kevin Vigor <kvigor@fb.com> | 2017-01-04 17:03:03 -0800 |
| commit | 63403742f53ec59a6acbe26ff4c39bab1b0842ed (patch) | |
| tree | 2c35221629def5e553caa957152ca5db72e59c65 | |
| parent | 1e69c7ef96a92754d4394fe2cff8767d3b1809fa (diff) | |
Another shot at stablizing halo prove tests
Summary:
This time for sure!
Changeset 7a6ead5 introduced a bug which allowed halo to exceed the
halo-max-replicas setting; the setting is only honored once
halo-min-samples pings have been received. This is fixed by using
the global halo-enabled flag when enforcing halo-max-replicas.
The remainder of this changeset is enhanced debugs.
Test Plan:
prove tests/basic/halo*
Reviewers:
Subscribers:
Tasks:
Blame Revision:
Change-Id: I8231a6c66f2c620d99e3418ed0eca757d50f085a
Signed-off-by: Kevin Vigor <kvigor@fb.com>
Reviewed-on: http://review.gluster.org/16326
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Shreyas Siravara <sshreyas@fb.com>
Smoke: Gluster Build System <jenkins@build.gluster.org>
| -rw-r--r-- | xlators/cluster/afr/src/afr-common.c | 29 |
1 files changed, 24 insertions, 5 deletions
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c index d5002a2070b..6431432690d 100644 --- a/xlators/cluster/afr/src/afr-common.c +++ b/xlators/cluster/afr/src/afr-common.c @@ -4452,6 +4452,17 @@ find_worst_up_child (xlator_t *this) return worst_child; } +static const char *halo_state_str(int i) +{ + switch (i) { + case 0: return "DOWN"; + case 1: return "UP"; + } + + return "unknown"; +} + + static void dump_halo_states (xlator_t *this) { afr_private_t *priv = NULL; int i = -1; @@ -4463,14 +4474,12 @@ static void dump_halo_states (xlator_t *this) { gf_log (this->name, GF_LOG_DEBUG, "Child %d halo state: %s (N/A)", i, - priv->child_up[i] ? CHILD_UP_STR : - CHILD_DOWN_STR); + halo_state_str(priv->child_up[i])); } else { gf_log (this->name, GF_LOG_DEBUG, "Child %d halo state: %s (%"PRIi64" ms)", i, - priv->child_up[i] ? CHILD_UP_STR : - CHILD_DOWN_STR, + halo_state_str(priv->child_up[i]), priv->child_latency[i]); } } @@ -4510,6 +4519,16 @@ _afr_handle_ping_event (xlator_t *this, xlator_t *child_xlator, latency_samples, priv->halo_min_samples); } + gf_log (this->name, GF_LOG_DEBUG, + "ping: child %u (%s) latency %"PRIu64" ms (max %"PRIu64" ms)" + " up_count %d (min %d) enabled %s", + idx, child_xlator ? child_xlator->name : "<null>", + *child_latency_msec, + halo_max_latency_msec, + up_children, + priv->halo_min_replicas, + child_halo_enabled ? "true" : "false"); + /* * Case 1: This child's latency exceeds the maximum allowable * for this halo. @@ -4648,7 +4667,7 @@ _afr_handle_child_up_event (xlator_t *this, xlator_t *child_xlator, goto out; } } - if (child_halo_enabled == _gf_true && + if (priv->halo_enabled && up_children > priv->halo_max_replicas && !priv->shd.iamshd) { if (was_down == _gf_true) |