Bug
The last_run record in trove_log never updates its counters (files_indexed, files_skipped, files_errored, total_chunks) during an indexing run. They stay at 0 for the entire run, even though the actual index is growing.
Evidence
Run 27 indexed ~3,800 files (32,015 → 35,812 total) but the run record showed:
{
"files_found": 34876,
"files_indexed": 0,
"files_skipped": 0,
"files_errored": 0,
"total_chunks": 0
}This was consistent across multiple status checks over ~2 hours while last_indexed timestamps and total_files confirmed active indexing.
Additional Context
- Previous runs (15, 16, 17) did have correct counters, so this may be a regression
- Orphaned runs (4, 11, 12, 14, 26) also show 0 counters with
finished_at: null — unclear if those crashed before any progress or hit the same bug
- The
finished_at field also never gets set, so the run stays in status: running permanently
Expected Behavior
Run record should update files_indexed, files_skipped, files_errored, and total_chunks as indexing progresses, and set finished_at + status: completed when done.
Bug
The
last_runrecord introve_lognever updates its counters (files_indexed,files_skipped,files_errored,total_chunks) during an indexing run. They stay at 0 for the entire run, even though the actual index is growing.Evidence
Run 27 indexed ~3,800 files (32,015 → 35,812 total) but the run record showed:
{ "files_found": 34876, "files_indexed": 0, "files_skipped": 0, "files_errored": 0, "total_chunks": 0 }This was consistent across multiple status checks over ~2 hours while
last_indexedtimestamps andtotal_filesconfirmed active indexing.Additional Context
finished_at: null— unclear if those crashed before any progress or hit the same bugfinished_atfield also never gets set, so the run stays instatus: runningpermanentlyExpected Behavior
Run record should update
files_indexed,files_skipped,files_errored, andtotal_chunksas indexing progresses, and setfinished_at+status: completedwhen done.