[bug]: investigating some abnormal behavior on one of our Lightning nodes after upgrading to lnd v0.20.0-beta

[pranav-punjabi]

Pre-Submission Checklist

• I have searched the existing issues and believe this is a new bug.
• I am not asking a question about how to use lnd, but reporting a bug (otherwise open a discussion).

LND Version

lnd v0.20.0-beta

LND Configuration

lnd.conf

Backend Version

Bitcoin Core v23.0

Backend Configuration

btcd.conf

OS/Distribution

Ubuntu 20.04

Bug Details & Steps to Reproduce

Observed Events
We have noticed a few unusual events on the node recently. At this stage we are not sure if they are related, so listing them separately:

1. Throughput spike
   A few days ago we observed a sudden increase in network throughput on the node, roughly ~30x higher than the normal traffic.
2. CPU spike
   Today we observed CPU usage rising above 90% on the node. We are still investigating what might have triggered it. (in normal scenario its 2-3%).
3. Large number of GetNodeInfo RPC calls
   While reviewing the logs, we noticed a large number of repeated GetNodeInfo RPC requests.
   We counted approximately ~850 requests within ~33 seconds.
   We checked our internal services and scripts and none of them appear to be making these RPC calls, so the source is currently unclear.

Issue:

Due to these events our LND nodes became unresponsive multiple times in last few weeks.

Questions

• What would be the recommended way to debug or trace the origin of these RPC requests?
• How we can identify what is the reason behind cpu spike and throughput spike ?

Expected Behavior

.

Debug Information

I’ve attached the log file for the spike duration.
We observed two spikes during this period:

1. From 2026-03-16 18:15 to 2026-03-16 18:35 (UTC)
2. From 2026-03-16 18:36 to 2026-03-16 19:00 (UTC)

lnd_issue_logs.log

Environment

No response

[ziggie1984]

would it be possible to make a cpu profile during this period:

you need to first enable the profiler on a specific port and then fetch the profile (in the example I used port 6060 but you can use any port you like):

  1. Enable the pprof server in lnd.conf:
  [pprof]
  pprof.profile=6060
  Restart lnd once for this to take effect.

  2. When you observe high CPU, capture a profile:
  curl -o cpu.pprof "http://localhost:6060/debug/pprof/profile?seconds=30"
  Run this while the high load is happening — it will sample for 30 seconds then write the file. You can increase the duration if the spikes are short-lived and you want more data.

  3. Share the cpu.pprof file.

[ziggie1984]

Moreover can you update to LND 20.1 which had couple of bugfixes.

[akshit-modi]

@ziggie1984 , we need to rollout to development first, then we can merge it into the production. It will take 2 weeks atleast. Meanwhile we have enabled profiler.