haproxy misconfiguration after restarting VPC with cleanup

[weizhouapache]

Steps to reproduce the issue

• Create 1 VPC
• create 2 VPC tiers, and a VM on each VPC tier
• acquire 2 public IPs
• create LB rules to the VM in each tier

The VR has correct haproxy configuration

• restart VPC with cleanup

The new VR has haproxy configuration for LBs of only 1 tier

health check failing

ISSUE TYPE

• Bug Report

COMPONENT NAME

VPC

CLOUDSTACK VERSION

4.18/4.19

CONFIGURATION

OS / ENVIRONMENT

SUMMARY

STEPS TO REPRODUCE

EXPECTED RESULTS

ACTUAL RESULTS

[DaanHoogland]

just reviewed both solutions (@JoaoJandre ) and both seem to do the job
#8881 does it by overloading the call to create the LB commands
#8765 does it by checking which list to load from the DB

If I understand correctly this creates the list multiple times in #8765, while in #8881 it is created only once. I see no problem with either code but I think Wei has a point and we should go with the overloading solution.

Note that this is purely from a code analysis point of view. I have not had time to test either yet but have started building the envs to check them.

[DaanHoogland]

@JoaoJandre both tested and checked /etc/haproxy/haproxy.cfg to be exactly the same as before restart. (cc @weizhouapache @SadiJr )

less work then I thought.

[JoaoJandre]

@JoaoJandre both tested and checked /etc/haproxy/haproxy.cfg to be exactly the same as before restart. (cc @weizhouapache @SadiJr )

less work then I thought.

@DaanHoogland I've also just finished testing both solutions. Both fix #8745. However, #8765 does indeed produce more commands than necessary, while #8881 will only send one command for all LBs. Thus I think we should go with #8881.

[JoaoJandre]

Fixed by #8881