ARM compatible image

[morucci]

Users relying on new Mac books are unable to run the Monocle compose. Indeed the Monocle container image is only compatible with x86_64 / amd64 architecture.

This issue is the place to discuss solutions such as:

• Produce an ARM compatible image
• Document a workaround

[morucci]

It seems that a workaround solution could be to set export DOCKER_DEFAULT_PLATFORM=linux/amd64 to add a layer of emulation for Docker.

[kartikeyrajvaidya]

this didn't work for me

[morucci]

Unfortunately I don't have such platform to test it. Perhaps you could share the error you get ?

The other solution might be the full virtualization where you could start a compatible system then start Monocle.

[charlesdg]

No specific logs on the errors.

Message I get during the launch:

[+] Running 4/6
 ⠿ Network monocle_default                                                                                                                                Created      0.1s
 ⠿ Container monocle-elastic-1                                                                                                                            Started      1.0s
 ⠿ Container monocle-api-1                                                                                                                                Started      1.2s
 ⠸ api The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested                  0.0s
 ⠿ Container monocle-crawler-1                                                                                                                            Started      2.3s
 ⠹ crawler The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested              0.0s

And in the logs:

monocle-api-1 exited with code 137
monocle-crawler-1 exited with code 0
monocle-api-1 exited with code 0
monocle-crawler-1 exited with code 0
monocle-api-1 exited with code 0
monocle-crawler-1 exited with code 0
monocle-api-1 exited with code 137
monocle-crawler-1 exited with code 137

[JohnFlyIII]

Same issue here so a +1 from me.

M1 Mac user

I'm going to try and work on alternatives, maybe just host in AWS as the product looks like what I need.

same results and no logs to speak of.

[charlesdg]

I succeeded to make it work by

• adding more RAM (from 8 to 16gb to docker). The error with code 137 seems to not be present anymore (137 was a OOM error).
• adding ingest.geoip.downloader.enabled: false to elastic/environment entry in the docker-compose file.

[sourceful-karlson]

@charlesdg can you post your setup in PR or something?

[3B00D]

Found a solution by running Docker Desktop with Rosetta enabled (had the first idea from ChatGPT).
It's under Beta features in the latest docker desktop version as of today, but it seems to work.

[farkmarnum]

+1 to using Rosetta in Docker Desktop — this unblocked me

[wjamro]

I managed to run Monocle on M1 Mac using the tips above:

1. Install Rosetta -> softwareupdate --install-rosetta
2. Increase memory in Docker Desktop (v4.22.1) to 16GB
3. export DOCKER_DEFAULT_PLATFORM=linux/amd64

[MisterJimson]

Where would I set export DOCKER_DEFAULT_PLATFORM=linux/amd64?

[mrdaliri]

@MisterJimson you can run it in your terminal session before starting monocle (docker-compose up -d)

[dougch]

This is an x86 specific project... The flake calls out x86 in the devShell, and the upstream container has x86 binaries... when I try to start this in an Arm env, I get:

api-1      | exec /bin/monocle: exec format error
crawler-1  | exec /bin/monocle: exec format error

If you're on Linux and don't have Rosetta, another option might be, for those with KVM, is docker with QEMU, but otherwise, even building the container will take some work on Arm.

[TristanCacqueray]

Note that running monocle from source should work on ARM (see the toolchain supported platforms here). With cabal and ghc, you can follow https://github.com/change-metrics/monocle/blob/master/CONTRIBUTING.md#build-using-cabal