Skip to content

Troubleshoot live sessions

Problems that show up while a net is running, grouped by what's on your screen. Most of them are the system telling you something deliberately rather than failing.

"Reconnecting — catching up…"

Symptoms: The banner says catching up and the connection pill reads catching up.

Cause: Your connection dropped and reconnected. The client is replaying the events it missed from its last known seq.

Fix:

  1. Wait. The replay is automatic and usually takes a moment.
  2. Don't refresh — refreshing throws away the resume cursor and forces a full snapshot instead.

Verify: The pill returns to live and the roster matches what other operators see.

"Lost the server — showing last known roster"

Symptoms: The banner offers a resync and the pill reads out of sync.

Cause: The client can't reach the server at all. What you're looking at is the last state it folded, held deliberately rather than blanked.

Fix:

  1. Check your own network first.
  2. Use the resync the banner offers.
  3. If resync keeps failing, the instance is down. Nothing is lost — the event log is authoritative in Postgres and the roster rebuilds from it when the server returns.

Verify: The pill returns to live.

The net is paused and the roster is frozen

Symptoms: Everyone sees a slate net paused status. Writes are refused.

Cause: This is not a connection problem. Net control's heartbeat stopped, so NetRoll paused the session to keep the log consistent.

Fix:

  1. If you're the NCS and you have reconnected, resume control.
  2. If you're a co-owner or logger and the NCS isn't coming back, claim control. The session resumes under you.
  3. If nobody claims it, a background job closes the session after 15 minutes and produces its summary.

Verify: The paused status clears and check-ins commit again.

See hand off net control.

A row is read-only and says another operator is editing

Symptoms: You can't edit an entry; it shows someone else's callsign as the editor.

Cause: They hold the soft lock on that entry. Locks run on a roughly 15-second sliding lease renewed by activity.

Fix:

  1. Wait for them to finish. The lock releases on its own.
  2. If they disconnected, the lock releases immediately when their heartbeat drops — no waiting out the lease.

Verify: The row becomes editable.

Your edit was rejected with a conflict

Symptoms: Saving an entry returns HTTP 409 or a conflict message.

Cause: Two operators committed against the same version of that entry. NetRoll rejects the stale write rather than silently discarding someone's edit.

Fix:

  1. Re-open the entry to see the current server state.
  2. Re-apply your change on top of what's there.
  3. Save again.

Verify: The save succeeds and the entry shows both changes.

A check-in won't commit

Symptoms: Your entry stays pending, or the commit is refused.

Cause: Several distinct causes look the same on screen.

Fix:

  1. Check the connection pill. A pending entry that never settles means the write hasn't reached the server.
  2. Check whether the net is paused. Writes are suspended while it is.
  3. Check whether the session is closed. A closed session accepts no further mutations.
  4. If you're checking yourself in, confirm you have a verified email and a reserved callsign.
  5. If you're logging for someone else, confirm your role allows it — see roles and permissions.

Verify: The entry appears in the roster with the correct source badge.

Viewers see a different roster from yours

Symptoms: Your console shows an entry that other people don't.

Cause: Almost always the pending state working as designed. The client that made a change shows its own entry as pending immediately; viewers render authoritative events only, so nobody ever sees another client's unconfirmed write.

Fix:

  1. Wait for the entry to settle from pending to confirmed.
  2. If it never settles, treat it as a check-in that won't commit, above.

Verify: The entry is present for everyone with the same seq ordering.

Export is refused

Symptoms: Downloading CSV or ADIF returns HTTP 403.

Cause: Export is an NCS and owner capability. It's also only available on a closed session.

Fix:

  1. Confirm the session is closed.
  2. Confirm you hold the NCS or Owner role on it.
  3. If you need the log but not the role, ask an owner to configure delivery so summaries reach you automatically.

Verify: The download starts and its row count matches the on-screen summary.

Get help

File an issue at github.com/nkbooth/netroll/issues. Include the session's state at the time, what the connection pill read, and whether other viewers saw the same thing.