AbstractWatchManager first reconnect failed, but reconnect future never complete

[23335049]

kubernetes-client/httpclient-okhttp/src/main/java/io/fabric8/kubernetes/client/okhttp/OkHttpWebSocketImpl.java

Line 79 in a91b17c

response.close();

We use pod inform, sometime the pod event never received after network problem recovery.
WatchConnectionManager first reconnect failed, but reconnect future never complete.
This current thread is not setUncaughtExceptionHandler, Is it possible for something unusual to happen in this place or other place?

[shawkins]

@23335049 sorry for the delayed reply. Looking at the underlying source, the only place something could occur is https://github.com/square/okhttp/blob/okhttp_3.14.x/okhttp/src/main/java/okhttp3/Response.java#L288 - I don't think we've encountered that before.

Did you try patching the source to do something like Optional.ofNullable(response.body()).ifPresent(ResponceBody::close) instead, or using a finally block to complete the future?

Alternatively did you test your scenario with another http client to verify that it's an issue with okhttp?