Version 1.18.2.1
Release Date: September 14, 2026
Release Type: Stable
Previously published as: 1.18.17 (image tag staging-1.18.17)
OData paged-read, delta, retry, CSRF and EDMX configuration in the flow designer, and more precise end bounds for time-range filters in monitoring.
Frontend Server
Bug Fixes
- Time-range end bounds follow the precision they were written in: The end of a time range was widened according to the smallest non-zero field of its value, so an end written as 11:00 covered the whole hour and 00:00 the whole day. The end now covers exactly one unit of the precision it was written in: a date covers that day,
HH:mmthat minute andHH:mm:ssthat second. - Open-ended monitoring windows end at the moment of the request: A monitoring read without an end previously ran to the end of the current day. It now runs up to the moment it is requested, so a refresh moves the end forward. The System Logs viewer follows the same rules and no longer shortens an end written to the second down to its minute.
- Rejected Requests time filter in API Management: An end written as 10:59 kept only the records at exactly 10:59:00, and a request without an end had no upper bound. The end now covers the minute as written, and a missing end means the moment of the request.
Frontend Web
New Features
- Paged Read panel on the OData start node: For the GET/Query operation, the OData start node gains a Paged Read panel with Paging Mode (None, Server or Client), Page Size, Max Pages and delta tracking (Delta Field and Delta Field Type, with a hint that depends on the OData version). While paging is enabled, Content Type is locked to JSON. Client paging requires a positive Page Size and a deterministic order — at least one Query Sort row or a manual
$orderby— and the designer reports this before the flow is saved. - Concurrent Consumers field: The Paged Read panel carries a Concurrent Consumers field that sets how many pages the flow may process at once; empty or 0 means the default of 5. The field is shown only while paging is enabled, next to Max Pages.
- Max Retry field: For paged reads that use After Flow, a Max Retry field next to Retry Delay sets the total number of delivery attempts a page gets before it is marked ERROR and reading pauses. Left empty, a page is retried without limit.
- Fetch CSRF Token option: The OData connector node gains an opt-in checkbox for write operations that makes the connector fetch a CSRF token from the service before writing.
- Message Headers are sent with "Get Entities and Fields", and Resource Path can be typed: The entity lookup on both OData nodes now sends the node's Message Headers, so a service that protects
$metadatawith a header such as an API key can be read. Resource Path is now a free-text field with suggestions: the fetched entity list remains available, and the path can also be typed when the lookup cannot run. - Entity list from an uploaded EDMX document: EDMX Resources joins the Resources panel (
.edmx/.xml), and the OData node can take its entity list from an uploaded document, the way the SOAP connector picks a WSDL. A checkbox selects one source at a time: ticked, the node offers an EDMX Resource selector whose entities fill Resource Path; unticked, it offers the live lookup. - HTTP Timeout grouped with the connection fields: On both the OData start node and the OData connector node, HTTP Timeout now sits with the address and connection fields instead of after the paging, delta and batch options.
Bug Fixes
- OData v4 nodes are always saved with the JSON content type: The rule that OData v4 implies JSON was not applied to flows that were already saved, and because v4 does not show the Content Type field, a stored Atom value could not be corrected. The rule now takes effect in the panel and whenever a flow is saved, including imported, copied and AI-generated flows. On v2 and v3, Atom remains a valid choice and is left unchanged.
- Unchecking "Use Manual Query" clears the query text on the OData start node: The manual query text was kept invisibly after the option was unchecked and continued to be applied. It is now cleared.
- A credential is required for Basic and OAuth2 on the OData start node: The validation that requires a credential when Basic or OAuth2 authentication is selected did not run on the OData start node, so a flow could be saved without one. It is now enforced.
- Unticking a field clears its parents in the OData field tree: Unticking the last selected field under a parent left the parent ticked, so
$selectand the write body kept a parent path with no selected fields beneath it. Parents are now cleared when nothing under them remains selected, on both the start node and the connector node.
1.18.2.1 is a stable release. Previous: 1.18.2.0