When retrieving specific genomes' AMR metadata using a genome ID list (from #29), BV-BRC will fail to fetch all specified IDs with some degree of randomness. Automatic download retrying fails repeatedly to fetch these data unless the command itself fails and is restarted, at which point the dice roll again.
The same command can be submitted repeatedly until the user gets lucky and all genome metadata download successfully. It's not clear what is introducing this stochasticity to the API download process, but we will need to add safeguards to ensure downloads complete successfully, and if they don't, 1) return an informative error to a user, and 2) attempt a redownload of those data until the problem is resolved.
Additionally, when adding these safeguards, we should consider validation and safeguards on the accession ID entry process as well, as the download is very sensitive to things like whitespace.
When retrieving specific genomes' AMR metadata using a genome ID list (from #29), BV-BRC will fail to fetch all specified IDs with some degree of randomness. Automatic download retrying fails repeatedly to fetch these data unless the command itself fails and is restarted, at which point the dice roll again.
The same command can be submitted repeatedly until the user gets lucky and all genome metadata download successfully. It's not clear what is introducing this stochasticity to the API download process, but we will need to add safeguards to ensure downloads complete successfully, and if they don't, 1) return an informative error to a user, and 2) attempt a redownload of those data until the problem is resolved.
Additionally, when adding these safeguards, we should consider validation and safeguards on the accession ID entry process as well, as the download is very sensitive to things like whitespace.