From ee46fabdca3bcedf09f521d392df37c3440c0781 Mon Sep 17 00:00:00 2001 From: Slavi Pantaleev Date: Wed, 19 May 2021 08:39:28 +0300 Subject: [PATCH] Make waiting time for --tags=start configurable Fixes https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/1062 --- roles/matrix-common-after/defaults/main.yml | 16 ++++++++++++++++ roles/matrix-common-after/tasks/start.yml | 2 +- 2 files changed, 17 insertions(+), 1 deletion(-) create mode 100644 roles/matrix-common-after/defaults/main.yml diff --git a/roles/matrix-common-after/defaults/main.yml b/roles/matrix-common-after/defaults/main.yml new file mode 100644 index 00000000..8112191a --- /dev/null +++ b/roles/matrix-common-after/defaults/main.yml @@ -0,0 +1,16 @@ +# Specifies how long to wait between starting systemd services and checking if they're started. +# +# A too low value may lead to a failure, as services may not have enough time to start and potentially fail. +# +# A value higher than 30 seconds (or any multiple of that) may also not work well, because a failing systemd service +# auto-restarts after 30 seconds (`RestartSec=30` in systemd service files). +# Checking if a service is running right after it had potentially restarted in such a way will lead us to +# thinking it's running, while it's merely starting again (and likely to fail again, given that it already did once). +# +# All of the services we manage are also started sequentially, which in itself can take a long time. +# There may be a ~10 second (or even larger) interval between starting the first service and starting the last one. +# This makes it even harder to pick a correct value. Such a 10 second gap and a waiting time of 20 seconds will +# put us right at the "dangerous" 30-second mark. +# +# We can try to measure this gap and adjust our waiting time accordingly, but we currently don't. +matrix_common_after_systemd_service_start_wait_for_timeout_seconds: 15 diff --git a/roles/matrix-common-after/tasks/start.yml b/roles/matrix-common-after/tasks/start.yml index 300de94c..42cfe80c 100644 --- a/roles/matrix-common-after/tasks/start.yml +++ b/roles/matrix-common-after/tasks/start.yml @@ -30,7 +30,7 @@ # as we may run into systemd's automatic restart logic retrying the service. - name: Wait a bit, so that services can start (or fail) wait_for: - timeout: 15 + timeout: "{{ matrix_common_after_systemd_service_start_wait_for_timeout_seconds }}" delegate_to: 127.0.0.1 become: false