Details
-
Bug
-
Resolution: Unresolved
-
Major
-
2.5.1, 3.0, 4.0.0
-
Security Level: Public
-
Untriaged
-
Unknown
Description
In CBSE-1842, "Detected time forward jump (or too large erlang scheduling latency)" on master (198) led to it auto-failing over node 205.
Through code inspection, it is not clear how the time jump could cause the auto-failover. The issue is most likely not in ns_server code.
Aliaksey thinks we need to look deeper into Erlangs's distributed communication - probably it gets stuck due to time jump.