Blockdaemon has scheduled Canton node maintenance for November 19, publishing separate mainnet, testnet and devnet notices on November 16 at 18:23 UTC. The provider lists a 14:00-15:00 UTC maintenance window and warns of up to 15 minutes of downtime per affected node. This is advance notice for its infrastructure, not an announcement that the Canton network will stop.

The three environments are assigned different software versions: mainnet 0.10.0, testnet 0.10.1 and devnet 0.10.2. Operators who use more than one environment should treat those as separate deployment targets. A testnet result on one version does not automatically describe the mainnet configuration.

The mainnet release includes breaking changes

Canton's official Splice release notes classify 0.10.0 as containing breaking API, protocol and Daml model changes. Among the API changes, several Scan and Super Validator endpoints impose a maximum of 1,000 items and reject larger requests. Integrations that assumed a larger response need to revisit how they request data.

The notes also say downloading a participant's private keys through its administration API is disabled by default. This affects workflows such as identity dumps. That is a reason to review recovery procedures before maintenance, not a reason to disable a protection automatically.

An operator's acceptance check should match the service it actually runs. A dashboard loading successfully is useful, but it does not show that a reporting job, recovery procedure or application integration still behaves as expected.

Installation and protocol activation are separate

The release notes explain that Canton 3.7 supports protocol version 37, but adopting that protocol version requires a Logical Synchronizer Upgrade. Installing software and activating a network protocol change are therefore distinct steps. Blockdaemon's maintenance notice should not be presented as proof that every associated protocol feature has already been activated.

That distinction also helps explain why release notes and provider notices belong together. One describes the software's capabilities and compatibility changes; the other identifies a provider's intended deployment window. Neither alone answers every question an application team has about its production environment.

Prepare a service check, not just a restart check

Before the window, identify which endpoints and nodes your application uses, which environment each belongs to, and who owns the post-maintenance check. Record outstanding operations so that an interrupted response is not mistaken for a failed transaction.

After service resumes, inspect the relevant transaction or application record before retrying an operation whose result is unclear. This is especially useful when a client timed out while a backend continued processing. A retry policy should reflect the application's behavior rather than assume every timeout means nothing happened.

For readers reviewing the broader responsibilities, our wallet and custody guide separates control of assets from access through a service. Our transaction fee guide explains why a fresh submission also deserves a fresh cost check.

The next confirmation comes from execution

At this report's preparation, November 19 is a scheduled maintenance date. The completion notice, the deployed version and application-level checks will be the evidence that the work finished as intended. Blockdaemon's advance warning gives users time to prepare; it does not establish an outage that has already occurred.