Forum Replies Created

Viewing 1 replies (of 1 total)
  • Thread Starter benushersmith

    (@benushersmith)

    Thanks for the reply, below I’m gathering everything in one place. Most of this was in the original post, but here it is consolidated so the thread stands as a complete record.

    Environment: UpdraftPlus 1.26.5, WordPress 7.0.1, PHP 8.1.34, managed hosting (34SP), reproducible after a clean UpdraftPlus reinstall.

    What we established, with 34SP’s senior support doing the server-side work:

    • Job creation works fine. The options, logs and metadata are written to the database correctly. The crash is at the point the job starts or resumes.
    • Running the resume event directly reproduces it: wp cron event run updraft_backup_resume returns zend_mm_heap corrupted.
    • The same command with the Zend allocator disabled (USE_ZEND_ALLOC=0) returns the identical error.
    • With --debug, WP-CLI, WordPress and PHP all load correctly. The crash occurs only at resume.
    • The PHP error log shows nothing at the failure timestamp, so the process is dying at the C level rather than raising a catchable PHP error.
    • Object caching disabled: same result. Files-only and database-only jobs: same result.
    • Every other plugin’s scheduled events run through the same mechanism with no error. Only UpdraftPlus’s updraft_backup_resume triggers the crash.
    • UpdraftPlus runs normally on other sites on the same hosting platform, so it appears specific to this site’s environment, which carries some host-level customisation.

    On your suggestions: Xdebug is not running on this production environment. OPcache, and specifically OPcache Fast Shutdown, were not isolated separately from the tests above, so that is a reasonable avenue and may well be where the interaction sits. That said, this is a small charity site, the work has been pro bono, and I have now moved it to an alternative backup plugin that is running cleanly, so I won’t be pursuing the root cause further.

    I mainly wanted to close the loop and leave the detail here in case it helps your team, or another user on similar hosting who runs into zend_mm_heap corrupted. Thanks again for taking a look.

Viewing 1 replies (of 1 total)