XRP Ledger Fix Misses the Checkpoint

A bug fix for the XRP Ledger was pushed back after support from trusted validators slipped for about two hours, wiping out the uninterrupted approval streak the network requires before a rules change can take effect. CoinDesk reported that the patch, known as fixCleanup3_4_0, had been scheduled to activate at 09:34 UTC on Oct. 23.

The mechanics are unforgiving in a way that matters for anyone building on the ledger. The network needs support from more than 80% of its 35 default trusted validators — 29 servers — and that threshold must hold at every roughly 15-minute check for 14 consecutive days. According to CoinDesk, citing XRPL Dashboard, support fell to 26 validators at 12:54 UTC on Oct. 22. By 14:46 UTC it had returned to 29, but the countdown had to start again from zero. XRPLWin now lists activation at 14:46 UTC on Nov. 4.

That sounds procedural until you look at what the patch is meant to close. CoinDesk said the change addresses a loophole described in the xrpld 3.4.0 release notes: under certain conditions, an account that should only be able to burn an issuer’s tokens could create new ones instead. The same update also changes how lending code determines when a loan is late or in default.

What is established is the drop in counted support and the resulting delay. What is not established is why the count fell. CoinDesk noted that public trackers do not show which validators stopped voting, and that a validator can disappear from the count simply because the network did not hear from it at the checkpoint — for example, if a server was offline or restarting. In governance systems like this, a brief technical absence can have the same effect as active opposition.

The delay also keeps waiting any institutions watching Ripple’s proposed on-ledger lending system, which CoinDesk said is still not live. That code is meant for a setup in which lenders place XRP or other tokens into vaults and a broker lends them out for a fixed term. Support for the bug fix has since climbed to 31 of 35 validators, giving the second run a little more margin than the first had.