-
Notifications
You must be signed in to change notification settings - Fork 300
HAProxy 1.8 plans? #521
Comments
This should be a good starting point - https://fabianlee.org/2017/10/15/haproxy-zero-downtime-reloads-with-haproxy-1-8-on-ubuntu-14-04/ |
If you have a public docker image with it ready I can test it on the cluster I am running. |
|
As far as I can tell it is working fine, been running for about two hours now. I don't think I have any way to test the new cache, but reloading is working very nice. |
Just want to say that I have been running it since then with no issues. I'm not sure what mesosphere needs to have tested to be able to accept the merge. If they have anything that they need to be looked at to be able to accept this I can see what I can do. |
FYI: I've published both a Docker image ( I think the biggest concern is that users that have custom HAPROXY_HEAD may run into issues. I've added some very basic logic to handle this, but ideally you should update your HAPROXY_HEAD to remove the I've done some basic load testing and it doesn't seem to drop any packets, but nothing heavy enough to confirm that it improves the ZDD stuff (mostly because I haven't reproduced dropped packets on a reload yet). Any feedback would be welcome. |
I'll switch from using the docker image @vixns made to that one. I have had no issues myself running it. I was not able to notice any packets get lost from reloads. Have not done long test only say 10k request at a time at 10-20 rps with services going up and down. |
Also fyi - published Docker image |
I would love to know if the new 1.8 release for haproxy will be getting looked at soon.
If any help would be needed I could see what I could to, likely mostly only testing, not too good with python.
The text was updated successfully, but these errors were encountered: