Server never marked as down and never removed about memcached

[ozten]

I'm also doing some manual testing, running two memcached and taking the first one listed offline.

I've tried various settings for timeout, retries, retry, and remove... but memcache keeps choosing the dead instance. It doesn't remove it from the pool.

For fallback I'm trying:

new Memcached('127.0.0.1:11211', {failOverServers: ['127.0.0.1:11212']});

(with other options and setting up event listeners, etc)

With 0.1.5, I don't get any issues logged.

With 0.0.11 I do get

{"level":"error","message":"Connect did not give a server","timestamp":"2013-01-17T21:19:43.778Z"}
{"level":"error","message":"memcached raised an issue: {\"tokens\":[\"11211\",\"localhost\"],\"messages\":[{\"code\":\"ECONNREFUSED\",\"errno\":\"ECONNREFUSED\",\"syscall\":\"connect\"}],\"retries\":2,\"totalRetries\":0}","timestamp":"2013-01-17T21:19:43.782Z"}
{"level":"error","message":"Connect did not give a server","timestamp":"2013-01-17T21:20:03.994Z"}
{"level":"error","message":"memcached raised an issue: {\"tokens\":[\"11211\",\"localhost\"],\"messages\":[{\"code\":\"ECONNREFUSED\",\"errno\":\"ECONNREFUSED\",\"syscall\":\"connect\"},{\"code\":\"ECONNREFUSED\",\"errno\":\"ECONNREFUSED\",\"syscall\":\"connect\"}],\"retries\":1,\"totalRetries\":1}","timestamp":"2013-01-17T21:20:03.999Z"}

My expectation was for there to be a failure, the server to be removed, and eventually the library would try to add the server back in after reconnect milliseconds.

[3rd-Eden]

I put this issue under the 1.0 milestone that I have scheduled to release on the 30th

[mridgway]

Is there a new timeline for this?

[3rd-Eden]

@mridgway not really, I under estimated the work that was needed for a stable 1.0 I'm working on & off on it atm. Sorry that I cannot provide you with more specific details.

[lpalomero]

Hello, how is it going this issue?

[ianshward]

@3rd-Eden do you have a general idea of what the issue may be? I may look into this.

[ianshward]

This should be working now in master, fixed in #132 and a few other recent commits.

[ozten]

Hawt!