71% of Flagged Bitcoin Lightning Nodes Ran Affected Lines
A study found that 71% of 550 public Bitcoin Lightning LND nodes flagged on Aug. 11 still ran affected software lines on Oct. 13, despite patches being available for more than a year.
A study published Oct. 21 found that 71% of 550 public Bitcoin Lightning LND nodes flagged in an Aug. 11 security disclosure still ran affected software release lines on Oct. 13. Patches for the six vulnerabilities had been available for more than a year.
About 10% of the nodes upgraded between the disclosure and the final study snapshot. Another 108 nodes were no longer classified as live.
The patches were included in LND version 0.19.0, released 446 days before the Aug. 11 disclosure. LND is software used to operate nodes on the Lightning Network, which processes Bitcoin payments outside the main blockchain.
Saurabh Kumar of Bitcoin Data Labs compared 814 daily snapshots of the public Lightning network graph. The analysis used software features advertised by nodes to estimate their release lines. It could not identify the exact patch version installed on each node.
Across nine LND vulnerability disclosures, Kumar found no statistically detectable acceleration in upgrades. The result does not rule out smaller responses. Nodes that upgraded without broadcasting new announcements could still have appeared to run affected software.
The largest upgrade waves followed releases from Umbrel, a platform that packages Bitcoin node software. The three largest waves involved between 262 and 385 nodes. At least 87% of those nodes broadcast updated software features within 48 hours of an Umbrel release.
Umbrel distributed those versions 19 to 41 days after the related LND releases. The study could not identify which platforms individual nodes used, so it could not determine whether automatic updates produced the observed pattern.
Upgrade delays also occurred outside security disclosures. Nodes running older software when LND versions 0.19 through 0.21 became available took an estimated median of 88 days to move to newer release lines.
The affected release lines represented 14.9% of live LND nodes at the time of the August disclosure, excluding an identified managed fleet. Those nodes accounted for 6.4% of channel capacity.
The study did not establish whether every node on an affected release line was exploitable. Exposure depends on the vulnerability, the node’s configuration and whether its software version was accurately advertised. Kumar published the study’s code and aggregate data with the preprint.








