LRSPipe uses storage to save job status and progress. This enables resuming paused jobs and finding out the status of a job. In this section we will look at a few configurable options for how LRSPipe stores it's state.
Be default, LRSPipe uses the local filesystem for storage. It stores job data in individual files in a directory. Without specifying any options it will store these files in the directory /store at the root of where you run the job (presumably your unzipped LRSPipe release directory). You can change that location with the following argument:
bin/run.sh ... \
--file-store-dir "../desired-storage-dir" \
...The directory location specified will be the storage location for job files.
NOTE: If you have existing jobs and you change locations between runs LRSPipe will no longer be able to find them.
If you wish to instead store job details in a Redis server you can do so by specifying the Redis server connection URI below:
bin/run.sh ... \
--storage redis
--redis-uri URI \
...The URI should be in Redis format, which includes the following possible formats:
redis://HOST[:PORT][?db=DATABASE[&password=PASSWORD]]
redis://HOST[:PORT][?password=PASSWORD[&db=DATABASE]]
redis://[:PASSWORD@]HOST[:PORT][/DATABASE]
redis://[:PASSWORD@]HOST[:PORT][?db=DATABASE]
redis://HOST[:PORT]/DATABASE[?password=PASSWORD]
If you omit the URI it will default to redis://0.0.0.0:6379.
Additionally if you wish to provide a custom string such that all LRSPipe keys contain it as a prefix (e.g. in the case of a shared-use Redis server), you can do so with the --redis-prefix flag. If you do not include this argument, all related Redis keys will be prefixed with xapipe.
If you do not wish to store job status and progress at all, and would like LRSPipe to completely refresh on every restart, you can do so by specifying noop for the --storage flag. Keep in mind this will result in you not being able to resume a job at all and you will lose your progress any time the process is interrupted.