Closed goober closed 4 years ago
After upgrading to: kong: 1.14 kong-ingress-controller: 0.6.1
And set KONG_NGINX_WORKER_PROCESSES
to 1
The above error messages goes away. ~However, it seems that I hit another issue were I constantly get the the below timeout message~:
UPDATE The below error message is due to our firewall and after setting KONG_ANONYMOUS_REPORTS: off
the message goes away. However, the increasing memory consumption is still there.
2019/11/05 11:22:29 [notice] 27#0: *4456557 [lua] cache.lua:321: purge(): [DB cache] purging (local) cache, client: 127.0.0.1, server: kong_admin, request: "POST /config?check_hash=1 HTTP/1.1", host: "localhost:8444"
--
| 2019/11/05 11:22:59 [warn] 27#0: *4456557 [lua] reports.lua:70: log(): [reports] could not connect to TCP socket: timeout, client: 127.0.0.1, server: kong_admin, request: "POST /config?check_hash=1 HTTP/1.1", host: "localhost:8444"
And the memory consumption constantly grows aswell:
That reports
thing is possibly because Kong is blocked on udp
to internet. On most recent version that would be tcp
. You can turn it off: https://docs.konghq.com/1.4.x/configuration/#anonymous_reports
Edit: oh, you figured it out already! :-)
I also have this problem. How to solve the growing memory? version: 1.4.0
After upgrading to: kong: 1.14 kong-ingress-controller: 0.6.1
And set
KONG_NGINX_WORKER_PROCESSES
to1
The above error messages goes away. ~However, it seems that I hit another issue were I constantly get the the below timeout message~:
UPDATE The below error message is due to our firewall and after setting
KONG_ANONYMOUS_REPORTS: off
the message goes away. However, the increasing memory consumption is still there.2019/11/05 11:22:29 [notice] 27#0: *4456557 [lua] cache.lua:321: purge(): [DB cache] purging (local) cache, client: 127.0.0.1, server: kong_admin, request: "POST /config?check_hash=1 HTTP/1.1", host: "localhost:8444" -- | 2019/11/05 11:22:59 [warn] 27#0: *4456557 [lua] reports.lua:70: log(): [reports] could not connect to TCP socket: timeout, client: 127.0.0.1, server: kong_admin, request: "POST /config?check_hash=1 HTTP/1.1", host: "localhost:8444"
And the memory consumption constantly grows aswell:
Excuse me, do you know how to solve this problem
@chenjinxuan The maintainers are currently investigating this issue. https://discuss.konghq.com/t/possible-memory-leak-kong-1-4-kic-0-6-1/4800/6
Any update on this? This is preventing us from using the latest version of Kong in production.
Some testing is still needed but PR #5229 might be the solution to this!
Hi all, I got same issue with Kong 1.3 and ingress 0.6.0, installed with helm (stable/kong, 0.19.1), DB-less mode. i have upgraded kong version to 1.4 with the same chart version and defining memory limits in K8S the memory gets stable and errors have disappeared, if no limits the memory increases without control. limits 5GB with 4 workers and stable around 3.5-4GB.
I tested the Fix in out INT environment and it resolved the memory leak
Thank you for your feedback, @edgarcolque !! We have now merged #5229 by @zeeshen !
@goober Thank you for the report! If this problem persists in the next release, feel free to reopen!
Thank you for looking into this. I will wait for the next release and verify it. Great job!
@goober @edgarcolque We are running kong gateway v-3.3.1 with dbless mode and see the same memory issue where it keeps growing and it is not going down. Please suggest.
Summary
We have recently started to recognize error events in our kong-proxy (see logs below). First occurrence happens a couple of minutes after startup.
Additional Details & Logs
Output from
curl https://localhost:8444/status
Kong configuration:
You can also see that the used memory keeps going upwards: