Containers left in an incomplete status are not recovered
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
paunch |
In Progress
|
Undecided
|
Takashi Kajinami |
Bug Description
# This issue was initially reported in https:/
Currently paunch looks for the existing container and compare image id and CONFIG_HASH
(and some additional environments if needed) to determine whether it should recreate the container.
When deployment times out while starting containers, paunch can leave containers
in an incomplete status as described below.
- The container was recreated, registered to systemd, but was not enabled/started
- The container was recreated, registered to systemd, enabled, but was started
- The container was recreated, but was not registered to systemd.
In such situation paunch can't detect these incomplete containers and next deployment successfully runs
without any errors, leaving these containers without any fix.
We should ensure paunch registers systemd service and starts these containers properly.
The fix was proposed to stable/ussuri: https:/ /review. opendev. org/c/openstack /paunch/ +/821990