Auto-fees fail in v0.16.0

[Wired4ncer]

When trying to enable auto-fees in latest version, there is an error

"Client error: Error calling /lnrpc.Lightning/UpdateChannelPolicy"

Multiple litd users reported same issue in our telegram group.
Running latest lnd v0.20 bitcoind v0.30.0 and litd v0.16.0

Litd is running as remote to lnd on the same machine with ubuntu 24

[ViktorT-11]

I just want to double check for some initial troubleshooting:
Has the lnd node(s) been configured with rpcmiddleware.enable=true, as that's required to use AutoFees in remote mode?
https://docs.lightning.engineering/lightning-network-tools/lightning-terminal/run-litd#configure-lnd

[bitromortac]

How recent was the failure and does it persist (once you see the next run about three days after)? There will be some work to show more details regarding the error, but you got it from the terminal web user interface, right? Did you exclude any peers from AutoFees?

[Wired4ncer]

I just want to double check for some initial troubleshooting: Has the lnd node(s) been configured with rpcmiddleware.enable=true, as that's required to use AutoFees in remote mode? https://docs.lightning.engineering/lightning-network-tools/lightning-terminal/run-litd#configure-lnd

Yes rpcmiddleware.enable=true is in my lnd.conf

[Wired4ncer]

How recent was the failure and does it persist (once you see the next run about three days after)? There will be some work to show more details regarding the error, but you got it from the terminal web user interface, right? Did you exclude any peers from AutoFees?

I just installed lightning terminal to my migrated node yesterday. Once I tried to start auto-fees this error appeared and auto-fees did not enable at all.

[Wired4ncer]

I did just try to start auto-fees again just now and it seem to have enabled without the error. Maybe it just needed a bit of time?

[Wired4ncer]

It did run once, they I excluded some channels and now its in the state "waiting for the first run" for about an hour with the details of:

{
"session_info": {
"state": "ACTIVE",
"created_at": "1767968190",
"last_successful_conn_timestamp": "1767968194",
"last_conn_attempt_timestamp": "1767968192",
"unsuccessful_attempt_count": 0,
"features": {
"AutoFees": {
"name": "AutoFees",
"last_run_timestamp": "1767967486",
"state": "ACTIVE",
"status": {
"status": "NotYetAssessed",
"details": ""
},
"updated_at": "1767968194",
"feature_runs": {
"feature_run_details": [],
"pagination": {
"last_offset": "0",
"next_offset": "0",
"total_count": "0"
}

[dekdynamics]

I also have this problem on The Wall, as I've reported in Slack to @jbrill and @levmi

AutoFees was working fine for me, and unfortunately I cant tell exactly when it started failing. Maybe it was on litd 16, or maybe it happened after the silent update to lightning.terminal.engineering (when the Last Run stuff was added to the AutoFees UI.)

I have tried:

1. Disabling / Re-enabling AutoFees
2. Restarting litd
3. Generating a new pairing phrase
4. Many combinations of the above

Eventually, I always get back to "NotYetAssessed" and then this error:

curl -sk "https://autopilot.lightning.finance:12010/v1/autopilotserver/sessions/$(litcli autopilot list | jq -r '.sessions[] | select(.session_state == "STATE_CREATED" and (.label | contains("AutoFees"))) | .local_public_key' | xxd -r -p | base64 | tr '+/' '-_')" | jq
{
  "session_info": {
    "state": "ACTIVE",
    "created_at": "1767969653",
    "last_successful_conn_timestamp": "1767969706",
    "last_conn_attempt_timestamp": "1767969705",
    "unsuccessful_attempt_count": 0,
    "features": {
      "AutoFees": {
        "name": "AutoFees",
        "last_run_timestamp": "1767969706",
        "state": "ACTIVE",
        "status": {
          "status": "NextRunScheduled",
          "details": "Next approximate run time: 2026-01-12 14:41:00 +0000 UTC"
        },
        "updated_at": "1767969706",
        "feature_runs": {
          "feature_run_details": [
            {
              "run_at": "1767969706",
              "error_type": "CLIENT_ERROR",
              "details": "Error calling `/lnrpc.Lightning/ListChannels`"
            }
          ],
          "pagination": {
            "last_offset": "0",
            "next_offset": "1",
            "total_count": "1"
          }
        }
      }
    }
  }
}

[bitromortac]

It did run once, they I excluded some channels and now its in the state "waiting for the first run" for about an hour with the details of:

{ "session_info": { "state": "ACTIVE", "created_at": "1767968190", "last_successful_conn_timestamp": "1767968194", "last_conn_attempt_timestamp": "1767968192", "unsuccessful_attempt_count": 0, "features": { "AutoFees": { "name": "AutoFees", "last_run_timestamp": "1767967486", "state": "ACTIVE", "status": { "status": "NotYetAssessed", "details": "" }, "updated_at": "1767968194", "feature_runs": { "feature_run_details": [], "pagination": { "last_offset": "0", "next_offset": "0", "total_count": "0" }

The cadence is three days unless you disable/re-enable the session (which is not recommended). We identified a problem with respect to channel exclusion, which will be fixed very soon. At the moment this only runs if no channels were excluded.

[bitromortac]

Thank you for the report @dekdynamics. This is likely a different issue, the autopilot probably can't make a connection to your node. I opened another issue here: #1197