aerospike_scan_*_async function parameter scan_id ignored

[jarni-ua]

How is that the variable scan_id is not initialized/set/utilized anywhere in the aerospike_scan_*_async functions?

For example: https://github.com/aerospike/aerospike-client-c/blob/master/src/main/aerospike/aerospike_scan.c#L1437

The function documentation doesn't mention that it is not implemented:

* @param scan_id		The id for the scan job.  Use NULL if the scan_id will not be used.

[BrianNichols]

The original scan implementation did not allow retries, so there was only one scan_id per scan. Around 5 years ago, the scan implementation was changed to "scan by partition" and allow those partition scans to be retryable. The server rejects duplicate scan_ids, so each retry had to use a different internally generated scan_id.

This made it impossible to use the legacy scan_id argument to aerospike_scan_async(). That scan_id has been ignored since that time. I will update the api doc to say that the scan_id argument is ignored.

[jarni-ua]

@BrianNichols Thank you very much for the explanation. Yes, documentation update would remove the confusion.

Two followup quesitons.

1. Is there a way to get status/monitor scan progress/wait for it to finish? Because of the scan_id issue I assume aerospike_scan_info and aerospike_scan_wait can't be used.
2. Is returning false from as_async_scan_listener the only way to interrupt scan?

[BrianNichols]

For foreground scans (like aerospike_scan_async()), the best that could be done is list all scan/query status:

asadm -e 'show jobs queries'

Background scans (aerospike_scan_background()) use the old scan protocol, so the scan_id can either be passed in or retrieved if a NULL scan_id is passed in. aerospike_scan_info() and aerospike_scan_wait() still work if the scan_id is known.

The scan can also be stopped via a server command:

asadm -e 'enable; manage jobs kill trid <scanid>'