opendev/system-config
Code Issues Proposed changes

Increase bridge's ansible ssh controlpersist timeout to 180s

Browse Source 
There is an Ansible bug where if successive tasks are separated in time
by the ssh controlpersist timeout Ansible will race ssh's updates to the
connection causing the second task to fail with an rc of -13 [0].

Statistically we believe that the longer times between tasks are less
likely. This means ssh controlpersist timeout values that are larger
will be less likely to have ansible hit this bug. Increase the value
from a default of 60s to 180s to take advantage of this probability.

[0] https://github.com/ansible/ansible/issues/81777

Change-Id: Ic40730c3e0bd814e6a5c739e4415657594362032
This commit is contained in:
Clark Boylan 2023-12-01 10:27:04 -08:00
parent 1d75147a4e
commit a45afbeac5
1 changed files with 4 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
playbooks/roles/install-ansible/templates/ansible.cfg.j2
View File

@ -29,6 +29,10 @@ any_unparsed_is_failed=True
[ssh_connection]
retries=3
pipelining = True
# Override the default ControlPersist value of 60s due to
# https://github.com/ansible/ansible/issues/81777 and the hunch that
# fewer task pairs have larger deltas between them.
ssh_args = -o ControlMaster=auto -o ControlPersist=180s
[callback_profile_tasks]
task_output_limit = 50