Details
-
Bug
-
Resolution: Fixed
-
Major
-
feature-backlog
-
None
-
Untriaged
-
Unknown
Description
The ingestion node has a fixed timeout threshold. There are two issues with that:
1. There is always an inverse DNS lookup from each ingestion node, which takes very long; (I'm not sure why such a inverse lookup is necessary and I have to hardcode the mapping in /etc/hosts to get this lookup done faster)
2. The timeout value is hardcoded and cannot be auto-adujusted. If any timeout exception happens, we end-up with partially ingested data and forever hanging feeds.