Installation and upgrades
Startup logs wait for changelog lock
The service keeps logging Waiting for changelog lock... or fails with Could not acquire change log lock. This happens when the database holds a stale Liquibase lock from an interrupted startup or shutdown.
Stop the Allure TestOps service.
Connect to the
testopsdatabase:psql -h 127.0.0.1 -U testops testopsCheck the lock state:
SELECT * FROM databasechangeloglock;Clear the stale lock:
UPDATE databasechangeloglock SET locked=FALSE, lockgranted=NULL, lockedby=NULL WHERE id=1;Start the service.
Upgrade fails with syntax error at or near "nulls"
Startup logs show a Liquibase failure with syntax error at or near "nulls", and the failed SQL statement contains nulls not distinct. This syntax is supported starting from PostgreSQL 15, which is the minimum version for Allure TestOps 5 and later.
Check the database version:
SELECT version();
Fix:
- If you already attempted the upgrade, restore the database from backup first.
- Upgrade PostgreSQL to version 15 or newer.
- Repeat the Allure TestOps upgrade.
UI actions fail after an upgrade
After an upgrade, actions that worked before can fail with 401, 403, 405, or 409 errors, or with a message that you do not have enough permissions. The browser still uses the cached frontend of the previous version, which calls the API in the old way. This can happen on both Server and Cloud.
Clear the site data in the browser:
- Sign out of Allure TestOps.
- Open the browser developer tools and go to the Application tab.
- Select Storage and click Clear site data.
- Reload the page while holding Shift.
In browsers without the Application tab, clear cookies and site data for the Allure TestOps address in the browser settings. Ask other users to do the same if they see the errors.
Integrations fail with unable to invoke cipher due to bad padding
Starting a CI job from Allure TestOps or another action that uses integration credentials fails, and the logs of the testops service contain unable to invoke cipher due to bad padding. Allure TestOps encrypts integration secrets with a crypto password, and the instance now runs with a different value than the one the secrets were encrypted with. This usually happens when the value was not carried over during a migration or redeployment, or was copied with extra characters, such as a trailing space.
Restore the original value in the new deployment configuration:
- Docker Compose:
TESTOPS_CRYPTO_PASSWORDin.env(ALLURE_CRYPTO_PASSWORDin version 4.x); - Kubernetes:
cryptoPassinvalues.yaml; - Linux packages:
ALLURE_CRYPTO_PASSWORDin/opt/testops/conf/testops.conf.
For the migration from version 4.x, see Copy encryption key from old configuration file to new one.
If the original value cannot be recovered, re-enter the affected integration credentials after the instance is running with the new crypto password.