Rename the tunable EventScriptBanCount to EventScriptTimeoutCount

since we no longer ban nodes when dodgy scripts continue to hang.

We now only mark nodes as unhealthy if monitor events fail or timeout. Never ban.

diff --git a/include/ctdb_private.h b/include/ctdb_private.h
index a5a931f1c5c80032e634ebb7d548012f014de3d7..e490b21a8cd42211a8806d870d3916210b9d1701 100644 (file)
--- a/include/ctdb_private.h
+++ b/include/ctdb_private.h
@@ -103,7 +103,7 @@ struct ctdb_tunable {
        uint32_t monitor_interval;
        uint32_t tickle_update_interval;
        uint32_t script_timeout;
-       uint32_t script_ban_count; /* ban after this many consec timeouts*/
+       uint32_t script_timeout_count; /* allow dodgy scripts to hang this many times in a row before we mark the node unhealthy */
        uint32_t recovery_grace_period;
        uint32_t recovery_ban_period;
        uint32_t database_hash_size;

diff --git a/server/ctdb_monitor.c b/server/ctdb_monitor.c
index 54cdac1b62527240bd164691fb2d3f37703f7e86..2bf5dcb99f5430c3c137af34dc53f863082301b2 100644 (file)
--- a/server/ctdb_monitor.c
+++ b/server/ctdb_monitor.c
@@ -126,8 +126,8 @@ static void ctdb_health_callback(struct ctdb_context *ctdb, int status, void *p)
        if (status == -ETIME) {
                ctdb->event_script_timeouts++;
 
-               if (ctdb->event_script_timeouts > ctdb->tunable.script_ban_count) {
-                       DEBUG(DEBUG_ERR, ("Maximum timeout count %u reached for eventscript. Making node unhealthy\n", ctdb->tunable.script_ban_count));
+               if (ctdb->event_script_timeouts >= ctdb->tunable.script_timeout_count) {
+                       DEBUG(DEBUG_ERR, ("Maximum timeout count %u reached for eventscript. Making node unhealthy\n", ctdb->tunable.script_timeout_count));
                } else {
                        /* We pretend this is OK. */
                        goto after_change_status;

diff --git a/server/ctdb_tunables.c b/server/ctdb_tunables.c
index e737e200dd397af3b38dce067da2a75ef2f3a1cc..872260384bdb1663eda9835ce85466d425caec96 100644 (file)
--- a/server/ctdb_tunables.c
+++ b/server/ctdb_tunables.c
@@ -38,7 +38,7 @@ static const struct {
        { "MonitorInterval",     15,  offsetof(struct ctdb_tunable, monitor_interval) },
        { "TickleUpdateInterval",20,  offsetof(struct ctdb_tunable, tickle_update_interval) },
        { "EventScriptTimeout",  30,  offsetof(struct ctdb_tunable, script_timeout) },
-       { "EventScriptBanCount", 10,  offsetof(struct ctdb_tunable, script_ban_count) },
+       { "EventScriptTimeoutCount", 1,  offsetof(struct ctdb_tunable, script_timeout_count) },
        { "RecoveryGracePeriod", 120,  offsetof(struct ctdb_tunable, recovery_grace_period) },
        { "RecoveryBanPeriod",  300,  offsetof(struct ctdb_tunable, recovery_ban_period) },
        { "DatabaseHashSize", 10000,  offsetof(struct ctdb_tunable, database_hash_size) },