Cross Cluster hangs frequently

(Trevan) #1

I've been trying to get cross cluster to work between two geographically remote clusters. I've gotten it to work but frequently, it just stops working. Querying _remote/info and remote indices hang indefinitely. But after some time (I don't know the length since I usually go do something else), it works again.

Is there some sort of logging that I can look at to see what is happening?

I'm running ES 5.6.2 in both clusters. Cluster A can talk to Cluster B on both the http transport and the es transport port. Cluster B can NOT talk to Cluster A, though. Only Cluster A is set up for cross cluster, though, so I believe that should not be an issue.

(Jason Tedor) #2

When it hangs, would you please take a stack dump on both clusters (use jstack) and share them?

(Trevan) #3

Which servers in the cluster should I run jstack on? In cluster A, we have almost 30 servers and the other has 3.

And do you want me to just paste them here?

(Jason Tedor) #4

Sorry for not being clear. The coordinating node in cluster A, and all nodes in cluster B (unless it's an outrageous number of nodes, in which case let us start with the gateway nodes in cluster B).

(Trevan) #5

I have the 4 files but I can't seem to upload them here. The upload appears to be limited to image files. How do you want me to share them?

Also, just to make sure I did it correctly, I waited for _remote/info to hang. I then did a jstack -F pid of the node on cluster A that I had queried _remote/info (I think that is what you mean by the coordinating node since I don't have special settings). I then ran jstack -F pid on the three nodes in cluster B.

(Jason Tedor) #6

Would you be willing to email them? My email address is my first name at

(Trevan) #7

I sent you them in an email.

(system) #8

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