You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello team, we have some servers running the haproxytech/haproxy-alpine Docker image version 3.0.1 (we also tested with version 3.0.2 with Data Plane API version 2.9.4) and we noticed that the memory consumption of the servers was only increasing, with the same number of registered sites, when analyzing the container we noticed that there were several "orphan" transactions in /tmp/haproxy that are created when we remove a site that does not exist ({"code":404,"message":"missing object: Site TestServer does not exists"}), after deleting these "orphaned" files from the container and restarting, memory consumption dropped drastically.
As a test, we created a new container and ran the API creating 1000 new sites with the example command below:
But if we run the same command above, the error message that the Site does not exist is returned, the "orphan" transaction file is created in /tmp/haproxy and memory consumption increases:
curl -X DELETE -u ubadmin:123456 -H "Content-Type: application/json" "http://127.0.0.1:5555/v2/services/haproxy/sites/TestServer?version=3"
{"code":404,"message":"missing object: Site TestServer does not exist"}
With each execution of the above command on a non-existent site, the container's memory consumption increases by approximately 40MB.
Thanks in advance.
The text was updated successfully, but these errors were encountered:
Hello team, we have some servers running the haproxytech/haproxy-alpine Docker image version 3.0.1 (we also tested with version 3.0.2 with Data Plane API version 2.9.4) and we noticed that the memory consumption of the servers was only increasing, with the same number of registered sites, when analyzing the container we noticed that there were several "orphan" transactions in /tmp/haproxy that are created when we remove a site that does not exist ({"code":404,"message":"missing object: Site TestServer does not exists"}), after deleting these "orphaned" files from the container and restarting, memory consumption dropped drastically.
As a test, we created a new container and ran the API creating 1000 new sites with the example command below:
curl -X POST -u ubadmin:123456 -H "Content-Type: application/json" -d '{"farms":[{"balance":{"algorithm":"roundrobin"},"mode":"tcp","name":"TestServer","servers":[{"address":"192.168.100.1","name":"TestServer","port":45397,"check":"disabled"}],"use_as":"default"}],"name":"TestServer","service":{"listeners":[{"address":"*","name":"TestServer","port":45397}],"mode":"tcp"}}' "http://127.0.0.1:5555/v2/services/haproxy/sites?version=1"
Then we delete the site and no "orphan" file is created in /tmp/haproxy:
curl -X DELETE -u ubadmin:123456 -H "Content-Type: application/json" "http://127.0.0.1:5555/v2/services/haproxy/sites/TestServer?version=2"
But if we run the same command above, the error message that the Site does not exist is returned, the "orphan" transaction file is created in /tmp/haproxy and memory consumption increases:
With each execution of the above command on a non-existent site, the container's memory consumption increases by approximately 40MB.
Thanks in advance.
The text was updated successfully, but these errors were encountered: