es-local-dev health check keeps failing

[yuyalun-allen]

log.txt

Hello dear developer. When I use this script to run es docker containers, the health check of es-local-dev keeps failing.

05:14:20 allen@desktop-allen Desktop → curl -fsSL https://elastic.co/start-local | sh

  ______ _           _   _
 |  ____| |         | | (_)
 | |__  | | __ _ ___| |_ _  ___
 |  __| | |/ _` / __| __| |/ __|
 | |____| | (_| \__ \ |_| | (__
 |______|_|\__,_|___/\__|_|\___|
-------------------------------------------------
🚀 Run Elasticsearch and Kibana for local testing
-------------------------------------------------

ℹ️  Do not use this script in a production environment

⌛️ Setting up Elasticsearch and Kibana v9.1.5...

- Generated random passwords
- Created the elastic-start-local folder containing the files:
  - .env, with settings
  - docker-compose.yml, for Docker services
  - start/stop/uninstall commands
- Running docker compose up --wait

[+] Running 6/6
 ✔ Network elastic-start-local_default           Created                                                                                                                 0.1s
 ✔ Volume elastic-start-local_dev-kibana         Created                                                                                                                 0.0s
 ✔ Volume elastic-start-local_dev-elasticsearch  Created                                                                                                                 0.0s
 ✘ Container es-local-dev                        Error                                                                                                                 584.9s
 ✔ Container kibana-local-settings               Created                                                                                                                 0.0s
 ✔ Container kibana-local-dev                    Created                                                                                                                 0.0s
dependency failed to start: container es-local-dev is unhealthy
Error: docker compose up --wait command failed!
sh: line 190: VERSION: unbound variable

But when I execute the health check command manually, it returns 200, which means the service started successfully.

05:27:36 allen@desktop-allen elastic-start-local → curl --silent --head --fail -u elastic:${ES_LOCAL_PASSWORD} http://localhost:9200
HTTP/1.1 200 OK
Content-Length: 540
Connection: keep-alive
Content-Type: application/json
Keep-Alive: timeout=4
Proxy-Connection: keep-alive
X-Elastic-Product: Elasticsearch

Can you help to figure this out?

[ezimuel]

Hi @yuyalun-allen , in the log that you shared I saw this warning:

{"@timestamp":"2025-10-10T09:15:09.166Z", "log.level": "WARN", "message":"initial heap size [134217728] 
not equal to maximum heap size [2147483648]; this can cause resize pauses; for more information see
[https://www.elastic.co/docs/deploy-manage/deploy/self-managed/bootstrap-checks?version=9.1#heap-size]", 
"ecs.version":

It seems, you don't have enough RAM memory, do you have at least 2 GB free? This can slow down all the process and potentially fail the es-local-dev healthy.
Can you also share the OS that you are using? Thanks!

[yuyalun-allen]

@ezimuel Thanks for your attention! I'm using archlinux natively, here is the output of neofetch

                   -`                    allen@matebook-allen 
                  .o+`                   -------------------- 
                 `ooo/                   OS: Arch Linux x86_64 
                `+oooo:                  Host: VGHH-XX M1010 
               `+oooooo:                 Kernel: 6.17.1-zen1-1-zen 
               -+oooooo+:                Uptime: 18 hours, 44 mins 
             `/:-:++oooo+:               Packages: 1513 (pacman) 
            `/++++/+++++++:              Shell: bash 5.3.3 
           `/++++++++++++++:             Resolution: 3120x2080 
          `/+++ooooooooooooo/`           DE: GNOME 49.0 
         ./ooosssso++osssssso+`          WM: Mutter 
        .oossssso-````/ossssss+`         WM Theme: Adwaita 
       -osssssso.      :ssssssso.        Theme: Adwaita [GTK2/3] 
      :osssssss/        osssso+++.       Icons: Adwaita [GTK2/3] 
     /ossssssss/        +ssssooo/-       Terminal: tmux 
   `/ossssso+/:-        -:/+osssso+-     CPU: Intel Ultra 7 155H (22) @ 1.400GHz 
  `+sso+:-`                 `.-/+oso:    GPU: Intel Arc Graphics] 
 `++:.                           `-/+/   Memory: 24415MiB / 31534MiB 

By the way, after setting up ES and Kibana manually, I found that this issue may be caused by http_proxy and https_proxy setting. When ES requested for a SSL certificate, the authenticated IP address was the IP address of virtual network card. So the https request heath check keeps failing. I solve this issue by reset http_proxy and https_proxy.

I suspect it was the same issue when setting up in docker. Considering my network environment, proxy almost always necessary when using docker (i.e. I set proxy both in $XDG_CONFIG/docker/config.json and /etc/systemd/system/docker.service.d/), what is the elegant solution to this docker setting up?

[ezimuel]

@yuyalun-allen I see, maybe you need to use the noProxy settings in your docker/cconfig.json file.

Moreover, we are not using https in start-local, we only use http with http://elasticsearch:9200 as internal docker endpoint and http://localhost:9200 from the host.

Maybe, you should set noProxy to elasticsearch,127.0.0.0/8 ?

[yuyalun-allen]

@ezimuel Oh, that's weird. If it's not the fault of HTTPS certificate, then what is going wrong in start-local? Should I configure something like vm memory for docker or java? Or is it something about domain name resolution?

When I changed the health check to use http://localhost:9200 before, it can pass the first container's health check. But it then failed when starting next kibana container.

[ezimuel]

@yuyalun-allen you don't have to change the docker-compose.yml file generated by start-local. The correct endpoint is http://elasticsearch:9200 inside the containers, do not use localhost. Can you try to uninstall start-local running the uninstall.sh script inside the elastic-start-local folder? After that, try to execute again the script as follows:

curl -fsSL https://elastic.co/start-local | sh

I would like to understand the issue from a clear setup. Thanks.

[yuyalun-allen]

@ezimuel I mean the original log and console output showed above is from a fresh installation. I just described different trials that I used to figure out this issue.

[ezimuel]

@yuyalun-allen ok, I'll double check the log. From a first review I didn't find any evidence of issue, apart from the warning that I already mentioned.

[ezimuel]

Dear @yuyalun-allen we just merged this PR that should fix the issue. Sorry for the long wait :-(. Let me know if this fixes the issue. You can test it using the following command:

curl -fsSL https://raw.githubusercontent.com/elastic/start-local/refs/heads/main/start-local.sh | sh

Thanks!

[yuyalun-allen]

Ok, thanks for your work! I used to setup this project in my college's DevOps course, and the course has finished now (I manually installed elastic search instead) . But I would try it anyway!