transient "agent is lost"

[Jason Hobbs]

Sometimes during deployment, the unit message is "agent is lost, sorry!", and then it goes away.

See ntp status here:
http://paste.ubuntu.com/21000924/

This is with 2.0 beta 13.

[Jason Hobbs (jason-hobbs)]

logs from the controller and machine the ntp/0 unit was on

[Anastasia (anastasia-macmood)]

Depending on context, load on the system and what you are doing, units will be connecting and re-connecting to api while the system stabilizes.

For example, when any changes are needed on running mongo configuration like enabling HA, when deploying units, provisioning machines, etc.

As long as the connection is established once these operations are complete, it sounds like Juju is running as intended.

If this behavior is causing issues, please provide the effects of the failure (why it is an issue) and a reproducible scenario.

[Anastasia (anastasia-macmood)]

Presence is being reworked and this bug will be addressed then.

[Tim Penhey (thumper)]

I believe this particular issue was resolved due to the new presence code in 2.6.