SEPE-1165: Fix three bugs that stop the agent working under a monitoring system

[eschoeller]

Stacked on #7. Found by installing the CI-built RPM on a real RHEL 8 cluster and driving it the way Naemon will — which is the only reason any of these surfaced. All three are inherited from upstream.

1. The agent segfaults when run as a non-root user

The whole point of this integration is Naemon invoking pdagent nagios enqueue as the naemon user. That crashed:

panic: runtime error: invalid memory address or nil pointer dereference
go.uber.org/zap.(*Logger).Sugar(...)
github.com/PagerDuty/go-pdagent/pkg/common.init.2()
        pkg/common/logging.go:22

init() discarded the error from Build() and called Sugar() on the result:

BaseLogger, _ = config.Build()   // error thrown away
Logger = BaseLogger.Sugar()      // nil deref

The daemon owns /var/log/pdagent, but the CLI runs as whoever invoked it — a monitoring user has no business writing the daemon's log, so Build() fails and the binary dies during package init, before main(), with a stack trace instead of a diagnostic.

Now falls back to stderr, then to a no-op logger, so Sugar() always has a receiver. Failing to open a log file shouldn't stop an event being enqueued. Extracted to newBaseLogger() with the path as a variable so the failure is testable; four tests added.

2. The daemon can't restart

pdagent[10769]: remove /var/run/pdagent/pidfile: permission denied
pdagent[10865]: pidfile already exists

/var/run/pdagent was shipped in the package. /run is tmpfs, so it doesn't survive a reboot and a package operation recreates it root-owned while the daemon runs as User=pdagent. Unlinking needs write permission on the directory, so the daemon couldn't remove its own pidfile — and every subsequent start failed.

Dropped from the package; RuntimeDirectory=pdagent in the unit instead. systemd creates it with the right ownership before ExecStart and removes it on stop, taking the pidfile with it. A stale pidfile becomes structurally impossible.

3. Every upgrade leaves the service down

Error writing config: Config File "/etc/pdagent/config.yaml" Already Exists
Error in POSTIN scriptlet in rpm package oit-sepe-go-pdagent

pdagent init refuses to overwrite an existing config and exits non-zero. Package managers preserve /etc/pdagent/config.yaml across removal and upgrade, so under set -e the scriptlet aborted — before the service was enabled and started. Only initialise when no config is present.

Verification

On both RHEL 8 peers:

Check	Before	After
pdagent health as naemon, production	segfault	OK
/run/pdagent ownership	root:root	pdagent:pdagent
3 consecutive restarts	failed	active, active, active
upgrade over existing install	scriptlet aborted, daemon down	active

Full Go suite passes.

Upstream

All three exist upstream and are worth offering back. #1 in particular means the agent cannot be driven by any unprivileged process in production mode, which is its primary use case.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[eschoeller]

Superseded by the recovery branch. Merging the stack bottom-up with squash merges deleted the base branches these PRs were stacked on, and GitHub auto-closes a PR whose base disappears. #5's content survived by riding along in #6's rebase; #7's and #8's did not. Both branches were intact, so the missing five commits were cherry-picked onto main unchanged and reopened as a single PR.