Prior to this commit they would be marked as dead if they had no
currently running allocations -- even though they would spring back to
life (running) if the cluster state changed such that a new eval+alloc
was created.
Ideally DriverNetwork would be fully populated in Driver.Prestart, but
Docker doesn't assign the container's IP until you start the container.
However, it's important to setup the port env vars before calling
Driver.Start, so Prestart should populate that.
Since I was already fixing consul's tls handling in #2645 I decided to
update consul/api and pre-emptively fix our tls handling against the
newest consul/api behavior. consul/api's handling of http.Transports has
improved but would have broken how we handled tls (again).
This would have made for a nasty surprise the next time we updated
consul/api.
The way the copying was happening on the alloc_runner was by temporarily
setting the alloc.Job to nil, copying and then restoring it. This
created an issue in which when the alloc was shared (which it is in
server/client mode and between alloc_runner/task_runner) there were race
conditions that could create a panic.
Fixes https://github.com/hashicorp/nomad/issues/2605
The Vault API returns a nil secret and nil error when reading an object
that doesn't exist. The old code assumed an error would be returned and
thus will panic when trying to validate a non-existant role.
This PR removes deepcopying of the job attached to the allocation in the
alloc runner. This operation is called very often so removing reflect
from the code path and the potentially large number of mallocs need to
create a job reduced memory and cpu pressure.
Previously was interpolating the original task's services again.
Fixes#2180
Also fixes a slight memory leak in the new consul agent. Script check
handles weren't being deleted after cancellation.
