Health monitors

Http monitor does a get request against the back-end server essentially equal to making a web call.

 

Tcp monitor would be equivalent to a telnet check.

 

Both are diff accomplishing 2 diff things. It depends on case to case basis but I don't see adding both the monitors would cause any issues other than consuming extra resource on F5 if the scale of the deployment is very huge. You can specify monitors to check pool members I don't see any benefit monitoring just nodes. I usually disable the node monitors by default when building the F5's. The duration by default is interval 5 sec and will time out at 16th sec if no response is received from the pool member. Use the formula (3xn)+1, where n is the interval if you need to customize the intervals.

 

suthomas1,

You should know that http Get happens after a successful 3 way handshake. So why would one simply add tcp monitor again, which is again a 3way handshake...

Well it depends on how you are assigning both the monitors to the pool. If you use

http or tcp

together, it could be because you are sure that tcp would work always and not sure of the content check monitor, whether the server team has built the path or not, often these happens during the migration works. You just don't want the pool to go down because the service is Up, but not the content check path, so you put tcp monitor, but along with that you also add http/https monitor to know if it really passes content check. In long run, you would be removing tcp and keeping http alone.

But if you do

http and tcp

it's really of no use.

a) Is it better to have both http & tcp monitor on it or just one of them & why?

 

http monitor will check content(response code ex: 200) and tcp monitor will check the service(port ex:80) in the device.

 

By having both you can easily determine the issue , whether it is service or content. if issue with service both monitor will fail, else issue with content.

 

Note :http monitor will do same job what tcp monitor is doing.If you frequently peforming change in server enable both otherwise http monitor is enough.

 

b) What is the interval of this monitor running against the pool/node?

 

interval: how often in seconds to send a request timeout: how long to wait for a successful response before marking the member down

 

timeout = 3 x interval + 1

 

Set interval Based on response between server and F5.

 

c) will this monitor generate traffic to the end node while doing its monitor?

 

yes, it will.3whs and get request involved during monitoring