# Import Process Delete

This service runs through the reports of a specified service (normally
`import-process-bridge`) and assembles distinct ids of all non-fatal messages
matching certain criteria. It subsequently sends a message for each found id.
These messages are meant to be consumed by services which run deletetion
processes for certain endpoints (e.g. see [`record-deleter`](https://gitlab.
switch.ch/memoriav/memobase/services/postprocessing/record-deleter)).

There are a number of filter you can apply on the parsed reports:

- time-based (`--created-before`, `--created-after`)
- session id-based (`--session-filter`)
- record id-based (`--record-filter`)
- record set id-based (`--record-set-filter`)
- institution id-based (`--institution-filter`)

Furthermore you have to provide a `session-id` (this is used by subsequent
services to distinguish between deletion batches) and you can optionally go for
a dry run (`--dry-run`). Beware that the latter can be ignored by services
downstream if they choose to discard the respective message header.

Finally there are a couple of more or less static settings which are set by
environmental variables:

- `KAFKA_BOOTSTRAP_SERVERS`: a comma-separated list of Kafka brokers
- `CLIENT_ID`: name of the client. This is only used as the prefix, since every
  time the service is started a random id is generated
- `TOPIC_IN`: name of input topic
- `TOPIC_OUT`: name of output topic
- `POLL_TIMEOUT`: polling timeout in milliseconds (i.e. the time the Kafka
  consumer waits for new messages before sending an empty batch, which is the
  signal for the service to terminate). **set this value with care**, because a
  too low value could lead to "message loss", while a too high value means a
  higher latency.