Some questions remain unanswered:
1) Why the swapped virtual memory at node-1 increased from 13.5 to 19G by this night, if many connections was blocked? Were there more and more connections and messages incoming form anaware OpenStack services or we just can see a memory leak issue in beam process? How to debug this?
2) Why blocked connections persist instead of being closed?
Some questions remain unanswered:
1) Why the swapped virtual memory at node-1 increased from 13.5 to 19G by this night, if many connections was blocked? Were there more and more connections and messages incoming form anaware OpenStack services or we just can see a memory leak issue in beam process? How to debug this?
2) Why blocked connections persist instead of being closed?