feat(deploy): capture service startup journal after each restart

After both `Start cortex.service` and `Start neuron.service`, sleep 10s
and run `journalctl --unit <unit> -I --no-pager` to record the latest
invocation's log in the workflow output. Step is guarded by
`if: always()` so a failed start still leaves a usable trace.

infra-setup.sh now adds gitea_ci to the systemd-journal group during
user provisioning, so `journalctl` works without a sudoers entry.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
2026-06-02 16:48:56 +03:00
parent ea1fdf8aa6
commit a26bb9f04b
2 changed files with 25 additions and 0 deletions

View File

@@ -70,6 +70,16 @@ jobs:
sudo /usr/bin/systemctl daemon-reload
sudo /usr/bin/systemctl start cortex.service'
# Wait for the service to either come up or wedge, then capture
# the latest-invocation journal. Runs even on prior failure so a
# failed start step still leaves a usable record in the deploy log.
- name: Capture cortex.service startup journal
if: always()
run: |
sleep 10
ssh gitea_ci@hanzalova.internal \
'journalctl --unit cortex.service -I --no-pager'
deploy-neurons:
needs: [deploy-cortex]
runs-on: fedora-43
@@ -124,3 +134,13 @@ jobs:
ssh gitea_ci@${{ matrix.host }} '
sudo /usr/bin/systemctl daemon-reload
sudo /usr/bin/systemctl start neuron.service'
# Wait for the service to either come up or wedge, then capture
# the latest-invocation journal. Runs even on prior failure so a
# failed start step still leaves a usable record in the deploy log.
- name: Capture neuron.service startup journal
if: always()
run: |
sleep 10
ssh gitea_ci@${{ matrix.host }} \
'journalctl --unit neuron.service -I --no-pager'

View File

@@ -53,6 +53,11 @@ for host in "${cortex_host}" "${neuron_hosts[@]}"; do
# own fresh lookup.
sudo install -d -o gitea_ci -g gitea_ci -m 0700 \
/var/lib/gitea_ci/.ssh
# Grant journal read access so the deploy workflow can capture
# `journalctl -u <unit> -I` after a service start without
# needing a sudoers entry. Idempotent — usermod -aG on an
# already-member is a no-op.
sudo usermod -aG systemd-journal gitea_ci
'; then
echo " failed to provision gitea_ci — skipping ${host}"
continue