armadaplatform / armada Goto Github PK

when i run armada stop
it still shows in armada list
but ip is n/a and status crashed

service is based on microservice_python3

How is armada supposed to get in /opt? Shouldn't there be something like git clone https://github.com/armadaplatform/armada.git /opt/armada somewhere in install.sh?

$ bash <(curl -sL http://armada.sh/install)
...
$ armada --help
/usr/bin/python: Error while finding spec for 'armada_command.armada' (<class 'ImportError'>: No module named 'armada_command')

After upgrading to Armada 1.7.1 my deployment scripts started to behave in weird way. Before, when I ran armada restart -a my-example then all 3 instances of my-example were restarted as expected. Right now new instances are spawned, but old services are kept on list with crashed status. Sample armada list after restart:

my-example  10.0.39.253:32781  a536934eff73    passing  ['env:dev']  2016-10-31 08:04
my-example  10.0.55.241:32784  3d58a4f2d648    passing  ['env:dev']  2016-10-31 08:05
my-example  10.0.88.245:32778  6084956429f6    passing  ['env:dev']  2016-10-31 08:05
my-example  n/a                428e32379320    crashed  ['env:dev']  2016-10-12 11:52
my-example  n/a                4b651ed6e7cf    crashed  ['env:dev']  2016-10-12 11:47
my-example  n/a                e7fd1b9263cf    crashed  ['env:dev']  2016-10-12 13:19

armada stop command's output looks like this:

Stopping service f6f5078e0cb0...
Service f6f5078e0cb0 has been stopped.

This may lead to a confusion especially when armada stop is called with mistyped container id, and there is no easy way to tell which service has been accidentally stopped. To alleviate this problem, we should display stopped service name, along with container id.

Currently, an image squashed with armada build --squash consists of only one layer, which prevents docker pull/push/build commands from reusing shared layers between images.

Perhaps, after building an image, we should only squash the layers introduced by the service, i.e. squash only layers after: FROM base_image in Dockerfile. We would still have a lot less layers than without squashing, and retain docker cache reusing.

After upgrading to Armada 1.4.0 the command armada ssh no longer accepts short form of container ids. E.g.:

vagrant@ubuntu-14:/projects/gumball-finder$ armada list
Name            Address              ID            Status   Tags
armada          192.168.2.116:8900   349cb2f73ea2  passing  -
gumball-finder  192.168.2.116:4999   9a01dc151544  passing  -

vagrant@ubuntu-14:/projects/gumball-finder$ armada ssh 9a0
Command failed: IndexError - list index out of range

vagrant@ubuntu-14:/projects/gumball-finder$ armada ssh gumball-finder
Connecting to obiady.gd...
root@9a01dc151544:/#