Skip to content
Independent XRP reference
Our standards
← All XRP news

XRP Ledger

XRPL 3.5.0 Scheduled for December 3 With Slow-Close Fix

The planned XRPL release targets intermittent slow ledger closes. Operators need to verify the running version and measure results; the announced date does not establish a completed repair.

By
A detached gold fountain-pen nib and black feed rest beside a cobalt pen body with an empty threaded fitting on ivory stone.

XRPL 3.5.0 now has a scheduled release date

XRP Ledger Operations has scheduled XRPL version 3.5.0 for December 3, 2026, with improvements and fixes aimed at resolving intermittent slow ledger closes. The November 13 announcement names the next software milestone; it does not establish that operators have deployed the release or that its intended correction has succeeded.

The operations team says interim measures have already improved close times and that monitoring continues. U.Today independently reported the planned version, date and repair scope on November 13. The useful change since October’s configuration response is a named software delivery target. Describing the correction as permanent states the developers’ intention, not a result observed after deployment.

For infrastructure teams, the announcement creates a preparation window: identify the affected servers, reserve staff for review and decide what evidence will close the maintenance ticket. For XRP holders, it supplies a concrete reliability milestone to follow. Neither group needs to convert a scheduled maintenance release into a forecast of token demand, transaction capacity or investment returns. Those conclusions require different evidence.

Section sources[1][2]

A release date is not a network-wide installation event

The November 13 operations notice is a schedule, not a published set of final release notes. U.Today likewise reports 3.5.0 as an upcoming update, rather than a completed deployment. The eventual entry in the XRPL Foundation release register matters because it should identify the actual software and explain its changes. A calendar reminder cannot supply those details. [1, 2, 3]

Analysis: distinguish the date developers publish a package from the date a particular business puts it into service. A provider might review the release immediately but deploy during its own maintenance window. Another might complete installation before its customer-facing backend has restarted. These are hypothetical rollout states, not observations of any operator’s December plan. A status label should name the completed step instead of collapsing them into a single upgraded badge.

Analysis: separate the change owner from the service owner in the rollout record. The person selecting a release may not operate every system that depends on it. Give each owner an explicit acceptance condition and escalation contact. That prevents a successful maintenance ticket from silently becoming an assurance about another team’s application.

Section sources[1][2][3][4]

Verify the running xrpld version, then identify the machine

A software package on disk and the process serving requests are separate things to verify. The official server_info reference exposes build_version for the running server, and QuickNode independently documents the same field. Analysis: record that live response alongside the installation record instead of using either as a substitute for the other. The official installation guide remains the place to check the procedure for the relevant operating system. [4, 5, 6]

Analysis: an acceptance record for December’s rollout should connect the selected release to an identified server and a dated response from that running process. Include the environment and the restart record, so someone reviewing the change can distinguish a production node from a rehearsal machine. If a business uses multiple backends, one successful response should not be treated as an inventory of the entire service. Ask how the provider established coverage of the machines in scope.

This is particularly useful at a handover between an infrastructure contractor and an application owner. A package-install receipt may satisfy the contractor’s immediate task while leaving the application owner unable to tell which process serves requests. Requiring the running version and server identity in the handover makes the completion claim testable. It also gives support staff a precise question when an upgrade notice and observed behavior appear inconsistent.

Section sources[4][5][6]

A node rollout cannot isolate the whole XRPL consensus result

XRPL consensus is designed to let participants agree on the ledger’s state and transaction history. Official and QuickNode server_info references describe last_close.converge_time_s as the time spent reaching consensus on the most recently validated ledger. That is a particular measurement associated with a ledger, not a complete end-user response-time measurement. Preserve the ledger identifier when interpreting it. [5, 6, 7]

Analysis: this creates an important limit for a staged upgrade. Comparing an upgraded node with an older node can help expose local differences, but both participate in or observe the same network. They are not two independent XRP Ledgers. Matching consensus observations on those machines would not, by itself, isolate the effect of one machine’s new software on the network as a whole.

Consider a hypothetical provider that upgrades one backend and sees normal close behavior on both its upgraded and unchanged machines. The defensible conclusion is that the trial machine remained compatible with the observed network during that window. Proving the intended network correction needs a broader account of deployed software, observed conditions and subsequent close behavior. Conversely, an unhealthy trial machine can still reveal a local deployment problem even while the shared ledger progresses normally.

Section sources[5][6][7][8]

Measure occasional delays without hiding them in an average

Because the announced problem is intermittent, the observation method matters. Google’s monitoring guidance explains why averages can conceal a slow tail and why sampling resolution affects what operators can detect. Its separate release-engineering guidance recommends evaluating a limited deployment before proceeding more widely. These are general engineering practices, not evidence that Google tested XRPL 3.5.0. [8, 9]

Analysis: define the observation window and the delay criterion before judging the rollout. Preserve the distribution of observations and the count that exceeded that criterion, alongside total observations. A report of no slow samples has a different meaning after a brief quiet interval than after a representative workload. Record missing samples as missing; otherwise, a monitoring interruption during a restart could make the record look artificially clean.

For an operator’s own service, predefine the conditions that pause expansion and the person responsible for deciding what happens next. Evaluate local errors and synchronization alongside the network observations. A limited rollout can help contain a local regression, but it cannot promise that an occasional network condition has been eliminated forever. This article proposes an evaluation method; it reports no performance test or measured improvement for the unreleased version.

Section sources[1][8][9]

What managed API customers should request after December 3

Analysis for exchanges, wallet teams and payment businesses: ask for a short completion record tied to your service, with the release deployed, deployment window, backend coverage and observation period. Have the provider distinguish completed work from continuing monitoring. That turns a general software announcement into information your own operations team can use, without requiring customers to administer the provider’s nodes.

A useful hypothetical response could say that every backend serving a specified endpoint is running the selected release, that a defined observation window has finished, and that one remaining issue is still being investigated. A less useful response would simply repeat the release date. The difference is scope and evidence, not the confidence of the language. Running-version documentation and staged-release practice provide the basis for asking these questions. [5, 6, 8]

Uncertainty remains explicit: the announcement does not provide a complete root-cause explanation, a quantified improvement target or a guarantee against future slow closes. The next consequential evidence is the published 3.5.0 release and its deployment results. Until then, the December 3 schedule supports preparation, while the effectiveness of the intended fix remains to be demonstrated.

Section sources[1][2][5][6][8]

What to watch next

  • • December 3, 2026: whether the XRPL Foundation publishes version 3.5.0, or Operations issues a revised date.
  • • The release notes: the identified slow-close correction, its testing scope and any operator-specific upgrade instructions.
  • • Provider deployment records: the release actually running and coverage of the backends serving each customer endpoint.
  • • Post-deployment observations: the measurement window, delay criterion, sample count and missing-data record, rather than an average alone.
  • • Any official root-cause or follow-up report connecting the intended fix to observed network behavior.

Sources and verification

We prioritize primary records and label supporting coverage. Dates reflect each source’s publication record.

  1. [1]XRP Ledger Operations: version 3.5.0 release scheduleprimary
  2. [2]U.Today: XRP Ledger announces December 3 fixsupporting
  3. [3]XRPL Foundation: xrpld release registerprimaryUndated reference
  4. [4]XRPL documentation: update manually on Ubuntu or DebianprimaryUndated reference
  5. [5]XRPL documentation: server_info for xrpldprimaryUndated reference
  6. [6]QuickNode: server_info for rippled, updated June 11supporting
  7. [7]XRPL documentation: consensus protocolprimaryUndated reference
  8. [8]Google SRE Workbook: canarying releasessupportingUndated reference
  9. [9]Google SRE: monitoring distributed systemssupportingUndated reference