Closed michal-sankot closed 1 year ago
Hi, you need to make sure that the ARM-version of the container is used. If it pulls x86, Rosetta is ran, which doesn't implement some cpu opcodes - hence you get SIG_ILL.
ok, thanks 👍
it worked with:
docker run --platform linux/arm64/v8 -p 8529:8529 -e ARANGO_ROOT_PASSWORD=openSesame arangodb/arangodb:3.11.2
it may be handy to add it as a note to https://www.arangodb.com/download-major/docker/ so that others on Mx don't hit the same issue.
Was there any investigation done into why it's choosing linux/amd64
on a linux/arm64/v8
host? Docker shouldn't do that, and I've indeed verified that Docker pulls the correct thing on a native host. 😕 Is it possible your Docker VM was running in x86 mode or something somehow? Was this on Docker Desktop? A hand-maintained VM? There's a pretty serious bug here somewhere, and I think it's one worth pulling on. :sweat_smile:
@tianon this seemed to be a MacOS host; It has a Rossetta x86 emulation by default. To the MacOS user it seems to be more or less opaque which binary he's running - unless its x86 and demands AVX2 - which isn't implemented in Rosettta. So @michal-sankot would probably need to revalidate which docker etc. was in use on his machine.
hey, i can validate it. though i'm not strong with docker. so if you give me steps/commands to run, i happily fire them to check it.
i've listed defualt vars (on mac) and it sets default docker platform to linux/amd64
. so that's why it assumed platform to be amd64 while it's arm (M1 chip).
DOCKER_DEFAULT_PLATFORM=linux/amd64
can you file
the docker binary? is it ARM? or is it M1? or is it a fat binary?
it's ARM
/usr/local/bin/docker: Mach-O 64-bit executable arm64
😬 setting DOCKER_DEFAULT_PLATFORM
on your host in this way is very very extremely not recommended (I think it might not even be an officially documented environment variable for exactly that reason)
hmm, i was using intel mac before, so maybe it got carried over when migrating to m1 model 🤔
I tried to run arango docker image as indicated in tutorial
And it crashes.
Is there any way to make it work?
System Version: macOS 13.5 (22G74) Chip: Apple M1 Pro