Make things more clear using Celery workers

compose/rabbitmq/rabbitmq.conf

@@ -6,8 +6,8 @@ listeners.tcp.default = 5672
 management.tcp.port = 15672
 
 # Extend the consumer timeout (the default is 30 minutes) as otherwise workers get killed
-# that take longer to acknowledge a task. But in our case this is mostly the case as
-# other (maybe higher priortized) analysis tasks are before them in the queue or the
-# analysis task was scheduled.
+# that take longer to acknowledge a task. This timeout starts when a task is fetched
+# by a worker. A worker can fetch multiple tasks which can be configured by
+# CELERY_WORKER_PREFETCH_MULTIPLIER in our settings file.
 # https://www.rabbitmq.com/consumers.html#acknowledgement-timeout
-consumer_timeout = 604800000 # 1 weeks
+consumer_timeout = 86400000 # 24 hours

radis/settings/base.py

@@ -316,7 +316,8 @@
 CELERY_TASK_DEFAULT_PRIORITY = 5
 
 # Only non prefetched tasks can be sorted by their priority. So we prefetch
-# only one task at a time.
+# only one task for each availalbe child process. The number of child processes
+# can be set with the -c parameter when starting the worker.
 CELERY_WORKER_PREFETCH_MULTIPLIER = 1
 
 # Only acknowledge the Celery task when it was finished by the worker.