Automated test results upload
Results are marked as orphaned
Orphaned results appear when Allure TestOps cannot match an uploaded result to an existing test case. Common causes:
- The result contains an
ALLURE_ID(also known asAS_ID) that does not exist in Allure TestOps. - The result has neither an
ALLURE_IDnor a full-path selector. - The result was generated by a non-official Allure adapter that does not write the full path.
ALLURE_IDis set to-1orunknownon purpose, to make the result orphaned.
Fixes:
- Use only an existing Allure ID from Allure TestOps, not an arbitrary value. For example, create an empty manual test case and add its ID to the automated test code.
- If you do not use an explicit Allure ID, make sure the result contains a stable full-path identity from an official Allure adapter.
- Switch to an official Allure adapter and regenerate the results.
- When the test is ready to update its test case, replace
-1orunknownwith the real Allure ID from Allure TestOps.
Refactoring created duplicate automated test cases
Renaming packages, classes, or methods changes the identity Allure TestOps uses to match results to test cases. Without an explicit Allure ID, each rename creates a new test case.
Fix:
- Assign explicit Allure IDs to tests before renaming or restructuring them.
- Treat parameter-set changes as identity changes unless you stabilize the test with an Allure ID.
A rerun stays In progress and later becomes Unknown
The incoming result did not match the expected retry context. Common causes:
- parameter values changed between the original run and the rerun;
- mapped environment values changed;
- the rerun happened in a different job or pipeline context, for example in a downstream pipeline that does not receive all environment variables from the pipeline started by Allure TestOps.
Allure TestOps matches the incoming result with the expected one by the history ID of the result. The history ID depends on the test, its parameter values, and its environment values:
historyId = md5(hash, sort(values(parameters)), sort(environment))
Here hash identifies the test. The calculation is stable: the same test with the same parameter and environment values always gets the same history ID. If parameter or environment values change from run to run, for example because they contain hashes, dates, or times, the rerun gets a different history ID. Allure TestOps treats it as a new result and adds it to the launch, while the expected result stays In progress and becomes Unknown when the launch is closed.
Fix:
- Rerun from the same job and pipeline context when you expect a retry. If the pipeline started by Allure TestOps triggers other pipelines, pass the Allure TestOps context to them, see Aggregate test results from several pipelines to a single launch.
- Move dynamic values out of mapped environments and parameters when they do not define a true execution context.
- If your framework supports excluded parameters, use them for values that should not affect retry matching, for example in allure-junit5 or allure-playwright. Check the README of your adapter in the Allure Framework repositories.
See: Launch environments, Jobs
Selective execution runs all tests instead of the selected ones
The full chain must be compatible: Allure TestOps triggers the job in two-way mode, the runtime receives ALLURE_TESTPLAN_PATH, the Allure adapter reads testplan.json, and no build tool filter overrides the Allure TestOps plan.
Fix:
- Confirm the job has a Build Server configured and is enabled for running tests from Allure TestOps.
- Refresh the job after CI-side changes.
- Remove extra CLI or framework filters that override the Allure TestOps test plan.
Launch is empty or some files are skipped
Common causes:
- the uploader points to the wrong results directory;
- files were already present before the uploader started watching the directory.
Fix:
- Verify the path contains the expected result files.
- If files are copied into place before the uploader starts, use
allurectl uploadinstead of watch mode. - For plugin-based uploads, set
indexExistingFiles: truewhen files already exist before the agent starts. - Enable debug logging in allurectl with
--log debugto see the full HTTP exchange with the Allure TestOps API. See Enable debug logging. - Check application-side logs for upload errors. On self-hosted deployments, your administrator can review the logs directly. On Cloud, contact support and provide the job and project details.
See: allurectl troubleshooting
Uploads fail due to file size limits
allurectl reports Failed to upload batch (…). Reason: failed sending upload request: … err: POST : 0. To see the cause, repeat the upload with --log debug, see Enable debug logging. The debug output contains the server response:
502 Bad Gatewayon Allure TestOps Cloud: the request body exceeds the 100 MB limit. Reduce the batch size with--size(ALLURE_SIZE), for example--size 50, and check whether the results contain large attachments, such as videos. To avoid uploading them, use--ignore-passed-test-attachmentsor--exclude-files, see Skip attachments for passed tests and generally skip some files.413 Request Entity Too Largeon a self-hosted instance: the reverse proxy rejects the request. Increase the maximum request body size, see Reverse proxy parameters, or reduce the batch size with--size.
A single *-result.json file larger than 2,000,000 bytes will not be processed as a test result. Large upload bodies can also be blocked by reverse-proxy limits before Allure TestOps processes them.
Fix:
- Move oversized textual payloads from result files to attachments.
- On self-hosted deployments, review reverse-proxy body-size and timeout settings.
See: Network and HTTPS, Cloud limits and quotas
A skipped or setup-failed test creates a duplicate test case
In some Allure adapter versions, the Allure ID label is written late in the execution lifecycle. If the test is skipped or fails during setup before the metadata is emitted, the result arrives without the Allure ID and without parameters and creates a new test case.
Fix:
Keep the Allure adapter version current.
Where your framework supports it, place the Allure ID in metadata that is emitted before the test body starts. For example, in Playwright, put the ID into the test title instead of setting it inside the test, see ID in the Allure Report documentation:
import { test } from "@playwright/test"; test("Create an issue @allure.id:123", async ({ page, browserName }) => { test.skip(browserName === "firefox", "Not supported yet"); // ... });If you cannot change when the adapter writes the Allure ID, set the Allure ID to
-1in the result when the test fails during setup. The result stays in the launch with the failure details, but does not create a test case, see Results are marked as orphaned.