Agent shows offline but it’s running
Your local process, container, or IDE plugin still runs, but Dailybot marks the agent offline or flips between states. Heartbeats are time-sensitive messages that prove liveness. Missed intervals, network drops, silent crashes, or clock skew all make the dashboard disagree with what you see on your machine.
Quick check
- Heartbeat interval — Confirm the agent configuration matches the expected ping frequency; sleeping laptops miss ticks when closed.
- Outbound HTTPS — Ensure firewalls allow the agent to reach Dailybot API endpoints without MITM breakage.
- Logs — Scan the agent stdout for heartbeat errors or authentication failures right before the status flips.
- Clock — Sync system time with NTP; large drift causes tokens and timestamps to fail validation.
- Version — Update to the latest agent build; older clients had tighter timeout bugs.
Common causes and fixes
Heartbeat interval missed
Agents typically send a lightweight ping on a schedule. Closing the lid, suspending Docker, or pausing the VM stops pings even though the binary is still “running” from your perspective. Keep the host awake during work hours, or run the agent on a small always-on server. Increase the interval only if product settings allow and your SLA accepts slower failure detection.
Network issue
Corporate proxies, captive portals, or flaky Wi-Fi drop TLS mid-request. The agent may retry, but a burst of failures marks you offline. Test with curl or similar from the same host to the API base URL (follow your internal security process). Whitelist required domains and modern TLS ciphers. Split VPN tunnels sometimes route only browser traffic; route agent traffic on the tunnel or bypass correctly.
Agent crashed silently
Wrapper scripts can respawn children that die immediately. Watch process lists and exit codes. Enable debug logging if supported. On Windows, check Service recovery options; on systemd, inspect journalctl for rapid restart loops.
Clock drift
If the system clock jumps backward or sits minutes ahead, JWT validation and heartbeat ordering fail. Enable OS-level time sync. After fixing time, restart the agent so it fetches fresh credentials.
Credential rotation
Expired API keys or rotated org tokens cause successful-looking local processes that cannot authenticate heartbeats. Regenerate the agent token in Dailybot, update local config or secrets manager, and restart. Confirm the org ID in config matches the dashboard org.
If none of this worked
Before contacting support, gather:
- Agent type (CLI, IDE, server) and version string
- OS and whether the host sleeps
- Redacted log excerpt around the first “offline” transition
- Output of a connectivity test to the API if your policy allows
- Organization name and agent display name in the dashboard
- Steps you already tried from this article
Then contact Dailybot support from the Help or Contact options in the product or on the website.