KingDezigns — PLEX32 Timezone Fix & Remote Reboot Script Summary

The problem

The PLEX32 backup script ran at 10:00 PM instead of the scheduled 2:00 AM, indicating a scheduling problem. A separate gap was also identified: there was no way to remotely reboot PLEX32 when away from home if it became unresponsive.

Backup ran at 10 PM instead of 2 AM No remote reboot capability for PLEX32

The crontab entry on PLEX32 was correct (0 2 */3 * *), but the system timezone was set to Etc/UTC instead of America/New_York, causing all cron jobs to fire 4 hours early during EDT (UTC-4).

Diagnostic commands run & why

1 Checked backup log timestamps

ls -lt /var/log/plex32-backup/

Confirm the 10 PM run actually happened and check if it was recurring.

Showed Jun 19 ran at 23:17, Jun 20 at 02:03, Jun 22 at 02:09 — pattern inconsistent, not a simple crontab typo.

2 Verified crontab entry on PLEX32

sudo crontab -l

Rule out a bad cron expression as the cause.

Crontab was correct: 0 2 */3 * * — not the problem.

3 Checked system timezone

timedatectl

A 4-hour offset is the classic symptom of UTC vs EDT mismatch.

Time zone: Etc/UTC — confirmed root cause. Cron was firing at 2 AM UTC = 10 PM EDT.

4 Checked timezones on all other VLAN 50 servers

ssh rufusking@192.168.150.35 "timedatectl | grep 'Time zone'"   # NAS08
ssh rufusking@192.168.150.40 "timedatectl | grep 'Time zone'"   # NAS16
ssh rufusking@192.168.150.30 "timedatectl | grep 'Time zone'"   # HAS

If PLEX32 had the wrong timezone, other servers might too.

All other servers were already on America/New_York — PLEX32 was the only one affected.

Root cause — cron firing at wrong time

PLEX32's system timezone was set to Etc/UTC (the Debian default). All cron jobs fired at the correct UTC time, which is 4 hours ahead of Eastern Daylight Time, causing the 2 AM backup to run at 10 PM EDT.

⚠ Key clue

The backup log on Jun 19 showed 23:17 and Jun 20 showed 02:03 — the jump from late-night to early-morning suggested the timezone was corrected mid-cycle, but timedatectl showed it was still UTC. The log timestamps themselves are in UTC, so both runs were actually at 2 AM UTC.

-rw-r--r-- 1 root      root      1559 Jun 22 02:09 backup-20260622.log
-rw-rw-r-- 1 rufusking rufusking 3120 Jun 20 02:03 backup-20260620.log
-rw-rw-r-- 1 rufusking rufusking 5879 Jun 19 23:17 backup-20260619.log

The Jun 19 timestamp difference was due to the backup running at a slightly different interval that cycle — not a timezone change.

Resolution: Set the timezone to America/New_York with sudo timedatectl set-timezone America/New_York. No crontab changes needed — cron reads system time automatically.

New capability — PLEX32 remote reboot script

A remote reboot script was created on NAS08 to provide a way to reboot PLEX32 from outside the home network via OMV's Scheduled Jobs interface. The script was iteratively debugged through several test runs.

nc not installed on NAS08 — port check always failed docker compose down clears restart policy — containers didn't auto-start flat sleep 60 caused unnecessary delay in Plex detection

1 Verified SSH key from NAS08 root → PLEX32 already existed

sudo ssh-copy-id -i /root/.ssh/id_ed25519.pub rufusking@192.168.150.45
sudo ssh -i /root/.ssh/id_ed25519 rufusking@192.168.150.45 "echo OK"

Key was already trusted — ssh-copy-id skipped it, echo OK confirmed passwordless access.

2 Added reboot to PLEX32 sudoers

sudo visudo -f /etc/sudoers.d/rufusking-nopasswd
# Appended: /sbin/reboot

The script runs sudo reboot over SSH — must be passwordless or it hangs.

Full sudoers line: apt-get update, fail2ban-client status, fail2ban-client status sshd, /sbin/reboot

3 First test — containers didn't restart after reboot

docker ps
# Result: empty — no containers running

The script ran docker compose down before rebooting, which clears the unless-stopped restart policy. Nothing brought the containers back up.

Plex was down after reboot — had to manually run docker compose up -d.

4 Fixed — added docker compose up -d after PLEX32 comes back online

ssh ... "cd ~/docker/plex && docker compose up -d"

Containers started cleanly after subsequent reboots.

5 Port check always reported failure despite Plex being up

nc -z -w 5 192.168.150.45 32400; echo "nc exit: $?"
# Result: nc exit: 127 (command not found)

nc (netcat) is not installed on NAS08 — the check always returned failure regardless of Plex state.

Switched to curl HTTP check — confirmed Plex returns HTTP 302 on port 32400.

6 Replaced flat sleep 60 with polling loop

while [ $PLEX_ELAPSED -lt 90 ]; do
  HTTP_CODE=$(curl ... http://192.168.150.45:32400/web)
  [ "$HTTP_CODE" = "302" ] && break
  sleep 5; PLEX_ELAPSED=$((PLEX_ELAPSED + 5))
done

Plex was up within ~10s of containers starting but the script was waiting a full 60s before checking.

Plex confirmed responding after 10s on final test run.

Script created

Location (NAS08)

/usr/scripts/nas/plex32_reboot.sh

Log directory

/var/log/plex32-reboot/reboot-YYYYMMDD-HHMMSS.log  (90 day retention)

What it does

1. SSH to PLEX32 → docker compose down (graceful container stop)
2. SSH to PLEX32 → sudo reboot
3. Wait 30s then poll ping every 10s (max 3 min) until PLEX32 responds
4. SSH to PLEX32 → docker compose up -d
5. Poll curl http://192.168.150.45:32400/web every 5s (max 90s)
6. Log success (HTTP 302/200/301) or warning if timeout

OMV Scheduled Job

Tag: PLEX32 Remote Reboot
Command: /usr/scripts/nas/plex32_reboot.sh
Enabled: OFF — enable manually, run once, disable again

Final test run result

[23:40:39] PLEX32 Remote Reboot — Initiated
[23:40:45] Docker containers stopped cleanly.
[23:40:45] Reboot command sent. PLEX32 is going down.
[23:41:27] PLEX32 is back online (ping responded after ~40s).
[23:41:30] Docker containers started.
[23:41:40] Plex is responding (HTTP 302) after 10s — reboot successful.

Lessons learned

Lesson	Detail
Always set timezone on new installs	Debian/Ubuntu default is `Etc/UTC`. Run `sudo timedatectl set-timezone America/New_York` immediately after setup. Verify with `timedatectl \| grep "Time zone"`. A 4-hour cron offset is the telltale symptom.
`docker compose down` clears restart policy	Unlike a system reboot (which respects `restart: unless-stopped`), an explicit `compose down` removes containers entirely. Always follow a pre-reboot `compose down` with a post-boot `compose up -d`.
`nc` is not installed on NAS08	Use `curl -s -o /dev/null -w "%{http_code}"` for HTTP service checks from NAS08. Plex returns HTTP 302 on port 32400 — treat 200, 301, and 302 all as success.
Flat sleep before health checks wastes time	Replace `sleep N` before a readiness check with a polling loop. Check every 5 seconds, break on success, cap with a max timeout. Services often come up well before the sleep expires.
Verify deployed scripts after editing	Use `grep -n "key_string" /path/to/script` to confirm the correct version is on disk before testing. Copy errors and stale files caused multiple confusing test failures in this session.

Documentation updated

File	What changed
`server_plex32.md`	Added timezone section (was UTC, corrected to America/New_York). Updated sudoers entry to include `/sbin/reboot`. Updated AI summary with timezone note and NAS08→PLEX32 SSH key entry.
`server_nas08.md`	Added `plex32_reboot.sh` to OMV Scheduled Jobs table (disabled/manual). Added NAS08 root → PLEX32 SSH key note. Updated AI summary with remote reboot script details.