Backblaze B2 Cloud Storage Now Has S3 Compatible APIs • Hacker News Board

Re: Backblaze B2 Cloud Storage Now Has S3 Compatible APIs

#291

Post by zimpenfish » Thu, May 07, 2020, 11:41 AM UTC

atYevP wrote: post #288

zimpenfish wrote: post #287

Earlier quoted context omitted.

I tried s3cmd according to their blog post without success. Just kept complaining that the access key was invalid. Which is a shame because `s3cmd` is much easier to use than `b2`.

Yev here -> make sure you ping our b2feedback@backblaze.com address and let us know about that experience, we're writing it all down and keeping tabs on what's not working as intended.

Thanks for following up! I eventually got it working - I'd followed the example too closely and had a rogue `us-west-002` left in the config which broke things because I'm apparently on `us-west-000`. But I'll drop an email anyway because I can't see an easy way to see what region you're in other than visually parsing the endpoint URL.

Re: Backblaze B2 Cloud Storage Now Has S3 Compatible APIs

#292

Post by rossjudson » Fri, May 08, 2020, 7:25 PM UTC

atYevP wrote: post #289

rossjudson wrote: post #281

Earlier quoted context omitted.

What about 0.1gb, 0.5gb? I guess that's my question. Is this a minimum charge per download?

Yev here -> No there's no minimum charges for downloads, it's just $0.01/GB down and $0.005/GB/Mo for storage, but everything is calculated on a byte hour level :)

Thanks! That's where I was curious.

Re: Backblaze B2 Cloud Storage Now Has S3 Compatible APIs

#293

Post by neurostimulant » Thu, May 14, 2020, 11:18 PM UTC

jorams wrote: post #107

neurostimulant wrote: post #76

Earlier quoted context omitted.

I really wish the B2 client support uploading file from Unix pipe. It would be nice to be able to archive a huge directory into a tar.bz archive and directly pipe the result into the B2 client without having to save the archive into disk first. Currently I have to save the tar.bz archive to disk first before uploading to balance. Took several hours to do so (huge spinning disks array, not as fast as ssd), while uploa…

Can you use process substitution? b2 upload_file bucket The argument the command sees will be something like "/dev/fd/42", and the shell will provide the output of tar through that file.

So apparently process substitution doesn't work. The b2 client is probably trying to read file size or something, keep failing with "ERROR: Invalid upload source: /dev/fd/xx". Maybe their api requires knowing the filesize upfront instead of allowing "streaming" upload.