Error log

The add-on records its own errors at AdminCP > Logs > Thread Split errors. The log is separate from the standard server error log and is scoped to events that are specific to chain handling: failed splits, watcher-migration retries, stuck splits cleared by the cleanup cron.

Common entries

Entry	What it means
Split failed: <reason>	The splitter aborted partway through. The chain row stays consistent; the previous part is not closed unless the split actually completed.
Watcher migration retry <n>	A batch of watcher subscriptions failed to insert. The cron will retry up to a fixed number of times.
Stuck split cleared	The 15-minute Cleanup stuck splits cron found a split that began but never completed (e.g. the request died mid-way) and reset its state so the next trigger can try again.

Entries are kept indefinitely. Filter by date or message text using the standard log controls.

When to escalate

If a single chain produces repeated Split failed entries, the chain is in a state the splitter cannot recover from automatically. Inspect the chain at the database level (xf_mc_threadsplit_chain and xf_mc_threadsplit_chain_thread) and consider detaching the most recent part to let the chain continue without it.