Feed lock mechanism for SaaS Data Export
The SaaS Data Export extension uses a feed lock mechanism to prevent race conditions when multiple processes attempt to sync the same feed concurrently. This can happen, for example, when a cron-triggered resync overlaps with a manual saas:resync CLI call.
How the feed lock works
Each feed sync operation—whether triggered by a cron job or a manual saas:resync CLI call—follows the same sequence:
-
The process tries to acquire the feed lock. The lock attempt is non-blocking and returns immediately if the lock is already held by another process.
-
If the lock is not available, the [operation is skipped and logged].
No data is lost. The next cron run picks up pending changes after the current process completes.
-
If the lock is acquired, the process records its name and PID for diagnostic purposes, then runs the sync.
-
When the sync completes or fails, the lock is released unconditionally so the next scheduled cron job can proceed normally.
Only one sync operation can hold the feed lock at a time, regardless of whether it was started by cron or the CLI. The feed lock is implemented through Adobe Commerce’s LockManagerInterface. The default backend is MySQL, which uses GET_LOCK and RELEASE_LOCK functions. To configure a different lock provider, see Configure the lock provider.
Expected log messages
The following message in commerce-data-export.log is normal and does not indicate a problem:
{"feed":"products","operation":"partial sync","status":"operation skipped - process locked by \"full sync(1234)\"", ...}
This message appears when a cron-triggered partial sync attempts to run while a full reindex or saas:resync is already in progress. The skipped operation is not lost. Once the running process completes and releases the lock, the next cron execution picks up and syncs any pending changes.
commerce-data-export.log, see Review logs and troubleshoot.