I'm not quite sure what you mean by client id stored in the database, since the client information is associated with a particular IRIDA instance (and so must already be in the database before upload of data)?
For resuming uploads though, I believe the IRIDA uploader writes a JSON file in the upload directory with information on what files were uploaded/when it was completed.
PUTcommands in there too for putting metadata into the system
Critical|Highissues as reported by Grype. Note the only one which would have been the
easiestto exploit was excel parsing used by the metadata importer. (https://github.com/phac-nml/irida/releases/tag/21.09.2)