I’ve read in several places that warnings about the task processor queue reaching its limit are an indication of an overloaded server. However, I see these warnings also on servers with relatively little load, which have many queues configured.
I can easily reproduce this on my test server, which has about 40 SIP peers, and 2000 queues defined in queues.conf. On an otherwise idle system, if I change queues.conf, and issue a
queue reload all command on the CLI, I get several messages like:
WARNING: taskprocessor.c:913 taskprocessor_push: The ‘subm:ast_device_state_topic-00000002’ task processor queue reached 500 scheduled tasks again.
To be honest, this worries me a bit. On live servers we see a similar behaviour when calls are coming in (it looks like when you have many queues, even a single call triggers the taskprocessor warning).
Since these servers are otherwise idle, and don’t process much traffic, I’m not inclined to look at adding hardware. So, is there something else I could look at?
This is with Debian Buster, asterisk 1:16.2.1~dfsg-1+deb10u1
Thanks for any tips! Best regards, Roel