All I’m having issues writing to Couchbase. I can easily overwhelm the Couchbase connection with multiple writes. I found in the documentation about implementing a back off protocol but only want to do that as a last resort. As a work around I have specified this same url in the “hosts”:[127.0.0.1:8091, 127.0.0.1:8091,…,127.0.0.1:8091] in the bucket configuration. It appears that each entry results in a different connection. The Couchbase client load balances across them, thus increasing my write bandwidth. Unfortunately I cannot delete from this bucket. What is the best way to maximize the number of writes to a single Couchbase instance? I’m using the Node.js driver which uses the C driver…
I’m not clear on how you’re observing the limitation. Is it that you just can’t get more throughput, or is it that you’re seeing TMPFAIL responses from the cluster. Note that we’ve not yet done another release, but a couple of recent patches give us a big performance boost on a single client instance.
Could you clarify, are you seeing TMPFAIL, or just observing throughput?
Have you updated to the latest Server and Client?
You should see some improvement.
Also as Matt asked how do you observe the limitation.