Turbo Indexer Log, PID and Chunk Files
What each file in /tmp is, and when it goes away
Files and logs
Every session keeps its files in /tmp, named with a session key: the server's machine name,
or server and index joined by two underscores for a run with --index.
drush ost --server=my_solr_server # key: my_solr_server
drush ost --index=content_index # key: my_solr_server__content_indexWithout --server, the server part is the first enabled Solr server.
Everything a background run prints. Never removed by the indexer; each run adds to the end.
The process ID of the session's main process. Removed when the run ends or drush osstop stops it.
The process IDs of the current round's workers. Removed at the same moments.
The session's server and index, which drush oss reads. Removed at the same moments.
One chunk file per worker, holding the item IDs of its batch. The files go after each round; the folder stays.
Versions older than 2.5 used a single /tmp/opensolr_turbo_master.pid.
drush osstop stops a process it finds there and removes the file.
Handy commands
# Follow a run live
tail -f /tmp/opensolr_turbo_my_solr_server.log
# The last 100 lines
tail -n 100 /tmp/opensolr_turbo_my_solr_server.log
# Worker errors
grep 'OpenSolr Worker' /tmp/opensolr_turbo_my_solr_server.log
# Sessions that have a main process file
ls /tmp/opensolr_turbo_*_master.pidCleaning up
drush osstop removes the process ID, state and chunk files of the sessions it stops.
Logs and empty chunk folders stay. Delete them when nothing runs:
rm -f /tmp/opensolr_turbo_*.log
rm -rf /tmp/opensolr_chunks_*Reference
- Workers and batch size
- Troubleshooting
- Files and logs