bhouse-nexthop commented on PR #14131:
URL: https://github.com/apache/cloudstack/pull/14131#issuecomment-6953431139

   @RosiKyu thanks for the thorough test, and for the clear repro. You're right 
about the leftover rule. Fixed in 8d3ec34af9.
   
   **The leftover rule.** `createLoadBalancerRule` created the rule first and 
stored the per rule settings after it, so the timeout check ran too late. The 
command now checks `idletimeout` and `keepalivetimeout` before the rule is 
created. A refused create leaves no rule and no firewall rule, the same as 
`algorithm=foo`. A new unit test (`CreateLoadBalancerRuleCmdTest`) checks that 
a refused value never reaches `createPublicLoadBalancerRule`. Any rule already 
left behind by the old build still needs deleting by hand, as you did.
   
   **`idletimeout=99999999999`.** Taken up: the API now refuses a timeout above 
2147483647 ms, the largest HAProxy accepts. The error says so, so the user no 
longer gets only "Failed to update load balancer rule". The check covers both 
timeouts, on create and on update.
   
   **`idletimeout=0`.** I've left this one as it is. `0` means no timeout, the 
same as for the global `network.loadbalancer.haproxy.idle.timeout` from #12586. 
When the global is 0, the generator removes `timeout client` and `timeout 
server` from `defaults`, and `haproxy -c` gives the same "missing timeouts" 
warning. The per rule value matches that meaning. The warning is HAProxy's 
advice against unlimited timeouts, and the parameter description already says 
what 0 does.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to