You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We try to determine if using docker swarm in production (for us or our clients) is a viable solution on along term basis.
I mean actually we are deploying our apps with docker-compose or kubernetes but we lack a simple multi-node deployment between these solutions.
Using docker swarm seems the obvious solution to jump from a mono machine deployment with docker compose to something very close but just with the multi machine feature.
The problem is that it is very difficult to find if swarm is a living project or not, because it relies on the docker-compose and moby/moby projects, which have a lot of activity, but the swarm project itself show only a few tags and no releases:
So if we try to estimate the liveliness with the release frequency swarm seems dead.
It is sad as it may not be the case and the project itself is a beautiful one.
Is there something wrong when I try to estimate the activity of the swarm project just by looking at the releases in github ?
Should I look somewhere else to justify to my manager going to swarm for production is credible ?
Yeah, Docker Swarm development seems very slow. Especially sad to see that there are existing pull requests waiting (support for devices, privileged), but no one is taking care to merge them.
It is stated that new issues should be filed in regular moby repository, which is still alive:
If your bug report relates to Docker's swarm mode, please file it against the Moby repository instead
You sometimes see "Swarm" in the regular Docker release notes, so I assume code from here is included in regular Docker releases. There are just no explicit releases here.
Mirantis bought the Swarm assets from Docker, and they tell every year "Swarm is here to stay" (2022, 2024), but it appears only for their paying customers, as nothing is really happening here.
But it has an advantage, as it seems k8s has a lot of development, lots of breaking changes (related k8s discussion). You don’t have such issues with Docker Swarm 😉
Final note: we use it in production, it satisfies 90% of our requirements, and we don't want to mess with k8s.
Final wish: would be awesome to have swarm in a separate plugin like compose, to develop and improve it further.
A lot of swarm relevant code actually lives in moby/moby. Swarmkit is essentially just the scheduler repo. A lot of work happened in moby/moby recently especially in the networking area.
Hello,
We try to determine if using docker swarm in production (for us or our clients) is a viable solution on along term basis.
I mean actually we are deploying our apps with docker-compose or kubernetes but we lack a simple multi-node deployment between these solutions.
Using docker swarm seems the obvious solution to jump from a mono machine deployment with docker compose to something very close but just with the multi machine feature.
The problem is that it is very difficult to find if swarm is a living project or not, because it relies on the docker-compose and moby/moby projects, which have a lot of activity, but the swarm project itself show only a few tags and no releases:
So if we try to estimate the liveliness with the release frequency swarm seems dead.
It is sad as it may not be the case and the project itself is a beautiful one.
Is there something wrong when I try to estimate the activity of the swarm project just by looking at the releases in github ?
Should I look somewhere else to justify to my manager going to swarm for production is credible ?
Related (old) discussion here
The text was updated successfully, but these errors were encountered: