healthcheck.sh exits prematurely due to `((var++))` under set -e

[nathanschram]

Problem

scripts/healthcheck.sh aborts after the first pass or fail call, reporting only one check instead of all five.

$ scripts/healthcheck.sh --version 0.35.1rc5
Checking service: untether.service
---
OK: service is active
$ echo $?
1

Root cause

The script runs under set -euo pipefail. Both counter helpers use post-increment arithmetic:

pass() { echo "OK: $1"; ((CHECKS_PASSED++)); }
fail() { echo "FAIL: $1"; ((CHECKS_FAILED++)); }

((expr)) returns a non-zero exit status when expr evaluates to 0. Post-increment returns the old value, so the first call (when the counter is 0) exits 1 and set -e terminates the script before any further checks run.

Fix

Use explicit assignment, which always exits 0:

pass() { echo "OK: $1"; CHECKS_PASSED=$((CHECKS_PASSED + 1)); }
fail() { echo "FAIL: $1"; CHECKS_FAILED=$((CHECKS_FAILED + 1)); }

Impact

• Post-deploy healthchecks have been silently skipping the version/errors/Bot-API checks since the script was added.
• Exit code 1 was masking the missing checks — deploy scripts using the healthcheck would have flagged successful deploys as failed.

Affected files

• scripts/healthcheck.sh (lines 26–27)

[nathanschram]

Additional bug found while fixing

Once the set -e crash was fixed, the error-log check started producing a false FAIL: 1 ERROR-level log entries in last 60s on a clean system:

ERROR_COUNT=$(journalctl --user -u "$SERVICE" -S "-60s" --no-pager -p err 2>/dev/null | grep -c . || true)

journalctl prints -- No entries -- to stdout when there are no matches, which grep -c . counts as one line.

Fix

Filter out journalctl meta lines:

ERROR_COUNT=$(journalctl --user -u "$SERVICE" -S "-60s" --no-pager -p err 2>/dev/null | grep -vc '^-- ' || true)

Both fixes go in the same PR since the script is unusable without both.