Apply several tunings around Ironic Conductor DB and process to reduce observed load

on the conductor host

etc/kayobe/kolla/config/ironic/ironic-conductor.conf

+[DEFAULT]
+debug = True
+# Try to make direct deploy faster
+force_raw_images = False
+
+[agent]
+# Direct deploy HTTP image source
+image_download_source = http
+# Reduce load by only collecting logs on failure
+deploy_logs_collect = on_failure
+
 [pxe]
 image_cache_size = 122880
-image_cache_ttl = 10080
+image_cache_ttl = 100800
 
 [console]
 socat_address = {% raw %}{{ api_interface_address }}{% endraw %}
 
 [conductor]
 automated_clean=true
+# We have busy conductors failing to heartbeat
+# Default is 10 secs
+heartbeat_interval = 30
+heartbeat_timeout = 360
+sync_local_state_interval = 360
+
+# Normally this is 100, but we see greenlets thrashing
+# and hitting timeouts, rather than making progress
+workers_pool_size = 15
+
+# Reduce background load
+periodic_max_workers = 1
+sync_power_state_workers = 1
+power_failure_recovery_interval = 0
+
+# Stop power state sync
+sync_power_state_interval = 0
+# Stop checking for orphan allocations for now
+check_allocations_interval = 0
+
+# Wait much longer before provision timeout check, to reduce background load
+# The default is 60 seconds
+check_provision_state_interval = 240
+check_rescue_state_interval = 240
+
+[database]
+# Seeing too many inflight requests to the DB
+# Reduce the size of the pool
+# Default is 50
+max_overflow = 8
+# Default is 5
+max_pool_size = 2
+# Default is 3600
+connection_recycle_time = 60
 
 [deploy]
 shred_random_overwrite_iterations = 0
@@ -18,5 +63,9 @@ continue_if_disk_secure_erase_fails = true
 # switches.
 timeout = 300
 request_timeout = 300
-# get some debug info
+# Get some debug info
 collect_timing = true
+
+[glance]
+# Retry image download at least once if failure
+num_retries = 1