Issue 246 auth backend #862
Conversation
Increases default lease TTL 5s -> 10s. Also timeouts earlier so lease_keepalive has a longer window to run (3s instead of 0.5s), which should make it less likely that unnecessary leader elections happen.
Adds followers section to api docs.
Exit with status 1 if lost leadership without previous graceful shutdown, so that systemd with "restart on failure" restarts it.
Yields during MessageStore#delete so that LavinMQ does not freeze while deleting large queues.
…deleted_definitions (cloudamqp#858) Speeds up the should_compact_definitions_during_runtime spec by setting max_deleted_definitions=8 instead of the default 8192.
|
I have read the CLA Document and I hereby sign the CLA 2 out of 5 committers have signed the CLA. |
|
I have read the CLA Document and I hereby sign the CLA |
|
Awesome! |
If a msg file is a few bytes larger than expected, first? will throw an IndexError and close the queue. This changes that behavior to instead rescue that IndexError and try to move on to the next segment file. Same fix as for shift? in cloudamqp#671 fixes cloudamqp#669 (again 🙂)
Don't redeclare exchange in test/java/com/rabbitmq/client/test/functional/Tables.java.
* rescue and close queue if LavinMQ starts with faulty msg_store --------- Co-authored-by: Jon Börjesson <[email protected]>
WHAT is this pull request doing?
Allows ConnectionFactory to use a chain of responsibility to authenticate users via multiple configured providers. A cache management is implemented to avoid calls that are too slow in http
WIP :
Initialize auth_chain with config.
Find a solution so that external auth services return LavinMQ::User
HOW can this pull request be tested?
Added fake
auth_http_backendserver in helper.Cache and
auth_chainhave dedicated tests