Created attachment 833455 [details] Output from starting rabbitmq-server Honestly, I have no clue about rabbitmq. I've noticed this because the test suite of dev-python/pika tries to start the server, and it threw a huge backtrace immediately. I can reproduce when executing rabbitmq-server directly. I'm attaching the output. I'll also attach erl_crash.dump compressed.
Created attachment 833457 [details] erl_crash.dump.xz
A quick fix for this is to move/remove /var/lib/rabbitmq/mnesia. The problem seems to be that there is a migration step required. I assume that folks that actually run rabbitmq know how to do this, but for the rest of us that don't care about persisting the datastore this fix should be sufficient.
Nope, still crashing. Mind you, I'm only trying to run pika's test suite, and FWICS src_test() is using a fresh directory anyway.
I tried to reproduce this with a focus on the src_test of dev-python/pika. But for me that code just worked. I first installed net-misc/rabbitmq-server and ran a test in a shell. > T=/tmp/rmq > mkdir $T > export RABBITMQ_LOG_BASE="${T}/rabbitmq/log" > export RABBITMQ_MNESIA_BASE="${T}/rabbitmq/mnesia" > export RABBITMQ_LOGS="${T}/rabbitmq.log" > export RABBITMQ_PID_FILE="${T}/rabbitmq.pid" > export RABBITMQ_ENABLED_PLUGINS_FILE="${T}/rabbitmq/enabled_plugins" > /usr/libexec/rabbitmq/rabbitmq-server -p 5672:5672 The service came up as expected. It also worked in src_test() of pika. But in the end there was an import error on "nose" in that function. Which indicates a problem in the pika ebuild, not surprising because that ebuild is actually test restricted. So this bug here needs more info and a repro. The src_test of pika can not be used as reasoning, because it is in fact restricted and can not be done on a clean tree.
