Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Nov 08 20:26
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Nov 08 20:20
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 18 21:29
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 16 23:25
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 16 23:25
    GitLab | Jiri Vlasak pushed 3 commits to damn_deploy
  • Oct 16 23:08
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 16 23:01
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 16 22:38
    GitLab | Jiri Vlasak pushed 4 commits to damn_deploy
  • Oct 16 21:49
    GitLab | Jiri Vlasak pushed 4 commits to damn_deploy
  • Oct 06 21:33
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 01 20:59
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Oct 01 19:55
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 23:52
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 23:06
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 23:06
    GitLab | Jiri Vlasak pushed 3 commits to damn_deploy
  • Sep 30 22:52
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 22:52
    GitLab | Jiri Vlasak pushed 3 commits to damn_deploy
  • Sep 30 22:49
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 22:49
    GitLab | Jiri Vlasak pushed 1 commits to damn_deploy
  • Sep 30 22:49
    GitLab | Jiri Vlasak pushed 4 commits to damn_deploy
Jiri Vlasak
@qeef_gitlab
Can you test it again?
matrixbot
@matrixbot
@karl:pesso.al Yep.. I'll
Jiri Vlasak
@qeef_gitlab
:+1:
Jiri Vlasak
@qeef_gitlab
So the problem is still here. I will work on it tonight. May I ask you if you can test yet with only one worker? It's ugly workaround, but could help temporarily.
matrixbot
@matrixbot
@karl:pesso.al do you know how to limit cpu on docker-compose file v3?
Jiri Vlasak
@qeef_gitlab
It should work like setting MAX_WORKERS environment variable. I used it in server/Dockerfile of the damn deploy repository.
when you run the server from cli, there is --workers INTEGER for uvicorn command
matrixbot
@matrixbot
@karl:pesso.al great
@karl:pesso.al min_size is greater than max_size
Jiri Vlasak
@qeef_gitlab
I think that min_size is 10 by default as max_size is
in Docker file, the default value for MAX_WORKERS is set to 100. The greater make no sense, imho, as it's the same as the default for maximum db connections.
matrixbot
@matrixbot
@karl:pesso.al Si I just builded using the MAX_WORKERS=1
@karl:pesso.al and I got the min_size error
@karl:pesso.al the min_size is related to the number of max workers?
Jiri Vlasak
@qeef_gitlab
oh, ok, so it looks like you will need to set up the MIN_WORKERS too, sorry.
Good news is that I found out how to replicate the issue in my dev environment, so I will work on fix locally.
matrixbot
@matrixbot
@karl:pesso.al Ohhh great :D
Jiri Vlasak
@qeef_gitlab
I am building server docker image with
MIN_WORKERS=1
MAX_WORKERS=1
It looks good for now
yet waiting for finish
matrixbot
@matrixbot
@karl:pesso.al I builded with min and max set to 1 and still got the same error
Jiri Vlasak
@qeef_gitlab
ok, I see, the same with me now.
matrixbot
@matrixbot
@karl:pesso.al yeap
Jiri Vlasak
@qeef_gitlab
please, use
ENV WEB_CONCURRENCY 1
instead of MAX_WORKERS
the server image from deploy repository uses https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker
matrixbot
@matrixbot
@karl:pesso.al ok
matrixbot
@matrixbot
@karl:pesso.al I just tried with web-concurrency=1 and it's working :D
@karl:pesso.al I'll use that until you came with another solution
Jiri Vlasak
@qeef_gitlab
:+1:
cool
I am working on this concurrency fix
It's annoying to use only one worker :/
matrixbot
@matrixbot
@karl:pesso.al in fact Gustavo Soares (Gitter) said that it's not that bad 😀
Jiri Vlasak
@qeef_gitlab
:) in fact, I run the https://client.damn-project.org/ on $5 instance which has only 1 core :D
I did load testing on that instance with not so bad results.
matrixbot
@matrixbot
@karl:pesso.al Great! So no rush on fix that... its working fine :)
Jiri Vlasak
@qeef_gitlab
I am glad to hear that. Thanks for help with debugging!
I will let you know as soon as the fix is out.
Gustavo Soares
@gustavo22soares_gitlab

:) in fact, I run the https://client.damn-project.org/ on $5 instance which has only 1 core :D

Nice, Before we used 1 vcpu and 512mb

Jiri Vlasak
@qeef_gitlab
:)
Jiri Vlasak
@qeef_gitlab

Hi guys @karl:pesso.al @gustavo22soares_gitlab . I would like to let you know that I finally fixed multi-process authentication issue. (For details, see damn-project/damn_server@a4d77c9 .) I also tested it but of course I do appreciate the second eyes.

I am also very thankful for the feedback when running the damn server in multi-process environment. So, I would like to invite you to set at most 9 uvicorn workers. (IIRC, the default value of your new instance was 8 workers.) Thanks!

Gustavo Soares
@gustavo22soares_gitlab
Thanks!
matrixbot
@matrixbot

@karl:pesso.al > <@gitter_qeef_gitlab:matrix.org> Hi guys @karl:pesso.al @gustavo22soares_gitlab . I would like to let you know that I finally fixed multi-process authentication issue. (For details, see damn-project/damn_server@a4d77c9 .) I also tested it but of course I do appreciate the second eyes.

I am also very thankful for the feedback when running the damn server in multi-process environment. So, I would like to invite you to set at most 9 uvicorn workers. (IIRC, the default value of your new instance was 8 workers.) Thanks!

Ohh I totally forgot about your message. We'll try that tomorrow :)

Jiri Vlasak
@qeef_gitlab

Ohh I totally forgot about your message. We'll try that tomorrow :)

:+1: there is no rush :)

matrixbot
@matrixbot
@karl:pesso.al Hey Jiri Vlasak (Gitter) just tested here if the default 8 workers and it's working great :)
@karl:pesso.al Thanks!
Jiri Vlasak
@qeef_gitlab
Cool! Glad to hear that. Let me know if you found any other troubles :) Thanks!
matrixbot
@matrixbot
@karl:pesso.al Sure!