Create a Timeout Config to LoadBalancers
I am very frustrated with the operation of the LoadBalancer. I have web services that lead to 2 minutes to answer a call and load balancer timeout is only 30 seconds and can not be changed. This behavior of the load balancer should only redirect calls to the servers, does not reproduce the actual operation of the server timeout is configurable. I put my clients to use loadbalance and started to have problems. At least this limitation had to be explicit in the description of the service so I would avoid spending so much time trying to identify the problem.
The suggestion of a support is that I create my own load balancer not what I believe to be the best answer.
The timeout feature is currently available in the CLB API with a default of 30 seconds and a maximum configurable value of 120 seconds. For this reason, I am marking this feature as complete. However, if you feel that this feature should be available in the control panel or should have a different maximum timeout value, please create a new feedback item so that it can be independently tracked and prioritized.
I have included a link to the API documentation.
Robert Navarro commented
Looking at the API docs it looks like there is an upper limit of 2 minutes for a timeout....is there a way we can make this higher?
Nigel Street commented
I assume the control panel option will also be available for Load Balancers associated with a Cloud Server? If not, please add that to the requirements :-)
Can I setup unlimited timeouts now? Where can I read the documentation?
Great news! Any ETA for the control panel?
Michael Weichert commented
awesome! can this be utilized at all in Cloud Sites?
Marbot Florent commented
Yes it kills all main CMS usage, wordpress, drupal...
Cloudsites loses one of the last good point it still have...
Stephen Furnival commented
Much agreed, the 30 second timeout kills an otherwise great product.
No solution, only a second I'm afraid. I am having a similar issue, an (admittedly slow) page times out rather than waiting. This makes previously slow operations unusable. Not good.
I moved to the balancer service from my own balancer (heartbeat / haproxy) to reduce work, not increase it.