As a dotmesh dev team, we'd like to be able to easily connect to a dotmesh cluster with dm so we can diagnose broken dotmesh installs - including things like runners, which set dotmesh up so that it's not exposed to the host network.
Currently, to do this on a runner, you need to docker exec into a container attached to the right network, download dm from get.dotmesh.io/unstable/master/Linux/dm, chmod +x it, examine the dotmesh-server-inner container's environment to get the initial admin password, do a dm remote add to get at it... and then you're in. Until the container gets recreated.
However, it wouldn't be too hard to put the dm binary in the dotmesh-server image in root's $PATH, and have require_zfs.sh spit out a working configuration in ~root/.dotmesh/config so we can just exec in and run dm. This would save a fiddly series of steps every time we need to go in and have a look.
ACs
[ ] It's possible to, on a docker install with a dotmesh server running, type (for instance) docker exec -ti dotmesh-server dm list and get back a list of dots.
alaric-dotmesh
commented
4 years ago
As a dotmesh dev team, we'd like to be able to easily connect to a dotmesh cluster with
dmso we can diagnose broken dotmesh installs - including things like runners, which set dotmesh up so that it's not exposed to the host network.Currently, to do this on a runner, you need to
docker execinto a container attached to the right network, downloaddmfromget.dotmesh.io/unstable/master/Linux/dm,chmod +xit, examine thedotmesh-server-innercontainer's environment to get the initial admin password, do adm remote addto get at it... and then you're in. Until the container gets recreated.However, it wouldn't be too hard to put the
dmbinary in thedotmesh-serverimage in root's$PATH, and haverequire_zfs.shspit out a working configuration in~root/.dotmesh/configso we can justexecin and rundm. This would save a fiddly series of steps every time we need to go in and have a look.ACs
docker exec -ti dotmesh-server dm listand get back a list of dots.