v0.61.62
Agent-only release. Fixes a disk/inode leak on managed fleets.
Fixed
- Pre-update rollback snapshots are now reliably cleaned up (GH #226). Before each plugin, theme, or core update the agent captures a snapshot under
wp-content/uploads/wpmgr-snapshots/so a failed update can be rolled back. Previously these were only cleaned up on a WordPress cron event (which never fires on sites withDISABLE_WP_CRONor very little traffic) or at the start of the next update, so a site that ran a batch of updates once and then went quiet kept every snapshot forever, quietly consuming disk space and inodes (the reporter saw thousands of files per site across a fleet).- Cleanup now runs opportunistically on ordinary agent activity, with no dependency on WordPress cron.
- A snapshot whose update succeeded is reclaimed within about an hour, once it is safely past the rollback window.
- Snapshots already accumulated before this release are swept on the first request each site handles after upgrading.
- Snapshots are never removed while a rollback could still be needed (a fixed minimum retention protects the control plane's post-update health-probe window and the update-safety watchdog), and a snapshot left behind by a failed rollback is retained for a few days for manual recovery.
No change to update, backup, or rollback behavior.
Full Changelog: v0.61.61...v0.61.62