log but do not fail if memcached operations do not get queued in time#102
log but do not fail if memcached operations do not get queued in time#102
Conversation
There was a problem hiding this comment.
This technically means that at the limit, we could experience timeout * 2 delay, correct? I.e. it could wait on the OpQueue right until timeout, when it gets taken out and runs, only to wait until timeout on the actual operation. This is not necessarily a problem, just want to understand if these are additive or if the thing is intelligent about taking into account both waits.
There was a problem hiding this comment.
Correct, this would mean it could take up to 1 second for an operation to complete, given that it will wait up to 500ms to get the operation in the queue and then up to 500ms for it to complete. Before it would wait up to 10s to get a slot in the queue.
From reading the spymemcached source code, it appears opTimeout is only used when doing synchronous gets / sets. In our case we set that timeout explicitely on the asyncGet call. I'm just setting it globally here in case it happens to affect anything else I am not aware of.
log but do not fail if memcached operations do not get queued in time
* Implement schema APIs * Inspection fix
No description provided.