Hi @avsej,
I just want to mention that I no longer see the huge (~6000ms) opening times when using the couchbase protocol. I’m not sure why or what’s changed. However, that protocol takes definitely takes longer (around 30ms) compared to the http method (around 4ms). This is repeatable.
I have uploaded a zip here containing the following files:
strace_cb.txt
strace_http.txt
strace_cb8091.txt
ltrace_cb.txt
ltrace_http.txt (fails)
ltrace_cb8091.txt (fails)
The *_cb files open the cluster with “couchbase://localhost”. The *_http files open the cluster with http://localhost:8091. The *_cb8091 files open the cluster with “couchbase://localhost:8091”
The two failure (ltrace with localhost:8091) gave a timeout error when opening the cluster:
err opening bucket: { CouchbaseError: Client-Side timeout exceeded for operation. Inspect network conditions or increase the timeout
message: ‘Client-Side timeout exceeded for operation. Inspect network conditions or increase the timeout’,
code: 23 }
This was repeatable with those protocols. It didn’t happen when running ltrace with “couchbase://localhost”. I’m not sure how to increase the timeout when opening a bucket.
I hope you can glean some useful information from this. Good luck!