When the master node publishes the cluster state, if a certain slave node fails to respond consistently and no timeout is set, does it mean that it will keep waiting and cannot complete?

Elastic Stack Elasticsearch
1

I found that when the master node publishes the cluster state, in the second phase, it will wait for all slave nodes to respond (successfully or unsuccessfully) before proceeding with the subsequent process. If a request is made to a slave node without setting a timeout and the slave node is in a high latency state, does this mean that the master node will keep waiting indefinitely？@DavidTurner

2

does this mean that the master node will keep waiting indefinitely?

No, see e.g. o.e.c.c.CoordinatorTests#testUnresponsiveFollowerDetectedEventually and o.e.c.c.CoordinatorTests#testAckListenerReceivesNoAckFromHangingFollower.

It's rather rude to @mention folks who haven't already shown interest in a topic. Normally if pinged like this I will just mute the topic without answering. I am making an exception in this one case, but please don't do it again.

Also Elasticsearch has no concept of a "slave node". This terminology is considered offensive.

2 Likes
4

Okay, I will avoid this situation in the future，so sorry.

1 Like
5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.