RabbitMQ crash with bump_reduce_memory_use

Question

I am using RabbitMQ 3.7.3 on Erlang 20.2.2 deployed on a docker (image rabbitmq:3.7-management).

Memory is setup like this : Memory high watermark set to 6000 MiB (6291456000 bytes) of 8192 MiB (8589934592 bytes) total

Here is the crash report that I am getting on automatic restart of RabbitMQ :

CRASH REPORT Process <0.818.0> with 0 neighbours exited with reason: no function clause matching rabbit_priority_queue:handle_info(bump_reduce_memory_use, {state,rabbit_variable_queue,[{10,{vqstate,{0,{[],[]}},{0,{[],[]}},{delta,undefined,0,0,undefined},...}},...],...}) line 396 in gen_server2:terminate/3 line 1161

It seems to be due to messages posted to a queue setup like this filled with 500k+ messages :

Thanks for your help !

Answer 1

I filed this bug and opened these pull requests to fix this issue - 3.7.x PR, master PR. This fix will ship in RabbitMQ 3.7.4.

In the future, it would be preferable to discuss or report issues on the mailing list as the RabbitMQ core team monitors it daily.

Thanks for reporting this issue and for using RabbitMQ.