Q: What is NetScaler Surge queue?

A: A Surge queue is a path in the NetScaler appliance through which all client connections are sent, irrespective of the condition of the target service, such as service being loaded or service has reached the maximum connections state. When the number of requests to the servers is low, the connections are not observed in the Surge queue because the connections are sent to the servers quickly and the Surge queue build up is not observed.

Q: When connection is in Surge queue, is there a way to change the number of retries before giving up a connection (default is 7)?

A: No, this is as per design and it is not recommended to change the number of retries.

Q: What is the total maximum interval of 7 attempts of retransmit before NetScaler gives up on a connection? How long does the 7 retries take in total?

A: When there is a SYN without a response, the time is doubled for the retransmit and the time keeps doubling for every SYN without a response.

If you were to capture an nstrace for analysis then you can see the following retry pattern interval - 1 second, 2 seconds, 4 seconds, 8 seconds, 16 seconds, 32 seconds, 64 seconds and then a RST is sent. This works as per exponential back off algorithm.

Q: How many connections can NetScaler surge queue handle?

A: Surge queue is essentially a list of memory buffer thus there is no hard limit and it can go on building as far as there is memory in the connection pool (NSB/PCB). Till date there is no failover or crash grade issues observed with Surge queue.