is there a way to have the activity monitor record the max RAM usage over a period of time for a container?

or are there a set of commands I can run to obtain that info myself?

You could roll your own monitoring -- prometheus plus the "node_exporter" can probably do this for you. But if you're not already using them, and don't expect to benefit from other nice features that they have, it might not be worth the effort.

are the docs open source? wanted to update this command https://docs.orbstack.dev/install#docker-migration to

sh
orb docker migrate

hey all! I've been using orbstack pro for a long time now. I recently tried out the Docker MCP Toolkit and it's pretty awesome. I was wondering if we can do something similar within orbstack? I'd like to run all my MCP servers locally via orbstack if possible

Good night everyone... Is there an update on macOS restart not shutting down correctly all the running containers? I have more than 60 containers running all the time, and restarting macOS just quits Orbstack in a second. I'm pretty sure it should take more time for this amount of containers and I'm worried about potential database corruption.

https://github.com/orbstack/orbstack/issues/1897 upload diagnostics to help supprot this i think

fixed

good databases should be able to handle abrupt shutdowns without getting corrupted

Ok… fair enough, that leads me to a new question then… Should Orbstack properly shutdown containers when macOS is restarting?

I'd expect it to try. But MacOS won't wait forever for orbstack to shut down, and who knows how long the containers themselves will take to go away.

fwiw I added [

init: true

](https://docs.docker.com/reference/compose-file/services/#init) to one of my services, precisely so that it would shut down quickly when I hit Control-C

Ok, that seems legit. Thank you.

Btw... do we have a consensus about how to better migrate an Orbstack installation to another Mac? I know about the Migration Assisstant issue with the data.img file and the tar workaround. However, that seems risky for my lack of experience. Would you say a Time Machine migration should work for a more "automated" migration?

time machine doesn't like it.

tar

is recommended

Ok, thank you... I will try to use tar then. Now I'm wondering... What if the data.img file size was customizable? I mean, instead of setting it to 8 TB by default, everyone could set it to a size that won't cause trouble for their own migrations.

@Danny I was able to migrate the data.img file. I will leave it here in case anyone needs it. (Important: the destination filesystem must be APFS) I used

bsdtar -C ~/Library/Group\ Containers/HUAQ24HBR6.dev.orbstack/data -cf - data.img | bsdtar -C /Volumes/MyDrive -xf -

to copy the data.img sparse file to an external drive Then I tried to do the same process backwards (copying from the external drive to the new Mac internal ssd), but for some reason it was writing the whole 8.8 TB file (I could track what the command was doing by looking at the Activity Monitor -bytes written in disk-) to the point where the internal ssd became full and I started getting errors from other apps. To stop this I had to restart, because Ctrl-C didn't work and even deleting the newly created file didn't free up the space. At this point I asked ChatGPT and it gave me a couple of options that didn't work either (

cmp

and

rsync

which didn't work either because they were trying to write the whole 8.8 TB too). I also tried copying and pasting via Finder, but it immediatelly errored out saying the file was too big for the destination. At the end, I could do it with plain old

cp

command:

cp /Volumes/MyDrive/data.img ~/Library/Group\ Containers/HUAQ24HBR6.dev.orbstack/data/data.img

. I had to track the progress via Activity Monitor, because the command didn't exit normally, but it did copy the sparse file correctly. Maybe I made a mistake at some point of the process that resulted in this painful experience. I still believe the sparse file is useful, but it would be a great idea to be able to set its size, just like you can in Virtual Machines softwares. This way, you don't have to go through this hassle when you need to migrate.

Hi all, I am a new OrbStack user, hopefully I am not asking a bad question here... I am trying to run a pretty large build, the SONiC build to be precise. The target of the build is x86 and the build documentation says that the ideal build host is Ubuntu 22.04, so I generated a machine with these two constraints. I have been working from the [documentation on the site](https://github.com/sonic-net/sonic-buildimage/blob/202505/README.md) as well as [this blog](https://support.stordis.com/hc/en-us/articles/19994539796381-How-to-build-the-community-version-of-SONiC-Virtual-Switch-SONiC-VS-image) in case the former is out-of-date. The host is a base spec M3 Ultra (96GB). Is it reasonable to expect that OrbStack (1.11.3) can handle this? This is the first time I've built the project so I don't have a reference. I've fixed one build issue with the locale not being properly set, The issue now is errors such as "make[4]: * write jobserver: Bad file descriptor. Stop." and am not sure what to do with those. Not sure if I am reading it correctly, but Rosetta seems to have 2.25M FDs allowed per process and there are a ton of these processes open.

I don't suppose anyone has been able to get wg-easy (https://github.com/wg-easy/wg-easy) - wireguard server container - running with OrbStack? I can get the container running but then can't route any traffic. Just wondered if it was actually possible with OrbStack

never heard of it. Maybe you could use "tailscale", which afaik is meant to work in containers

I've never been able to run any wireguard solutions in orbstack, at some point I just gave up and use the one in my router. Also, in my case, hosting it in a machine under the router is just putting multiple single point of failure in the setup, either of them dies will kill the whole connection.

That's very useful to know I'm not missing something super obvious! Thanks very much for taking the time. I wonder what's causing the issue 🙂

The challenge I ran into is that currently all the networking flows through the host stack and isn’t like Linux or Windows where you get an actual standalone network stack and the container can show up properly on the network. It should be possible to stand up wireguard connections however will be limited and specific

Very interesting... sounds above my paygrade to get that happening. I have had success in the past with VMWare and a debian container with the Algo project (easy setup for Wireguard and IKEV2) on a macOS host machine. But I'm sure that's all quite different underpinnings

I've used vmware (debian 12) and macpine (alpine 3.something) in the past before orbstack, both of them were one take pass when setting up wireguard, wg-easy specifically.

I'd not heard of macpine at all - looks interesting. I wonder how it compares to UTM

Just an qemu alpine vm wrapper, it doesn't provide much functions but good enough for me

simple and straight forward is always a good thing

Yeah

Another obscure question - I don't suppose OrbStack would have some timeout in the network stack? I have a container with a long running TCP connection - I'm setting up Z-Push, an activesync open source clone, which has very long running TCP connections (like up to an hour) and when I run it on native docker on a Raspberry Pi I don't have a timeout issue but on OrbStack the TCP connection seems to get killed after approx 20 mins. I'm trying to track down what's killing the connection and was wondering if OrbStack might be doing something like the official docker on Mac does with 'slip/port-max-idle-timeout' or 'vpnKitMaxPortIdleTime' ?