I succeeded in using the droplet interface to backup to AWS
use_https = false
backend = s3
aws_auth_sign_version = 4
and to our local ceph system
use_https = true
backend = s3
aws_auth_sign_version = 4
[For whatever it may be worth, the local ceph backup was very unstable on writing, and therefore failed on restoration when it hit one of the ubiquitious empty objects. I suspect our ceph configuration is at fault, since bareos works with amazon.]
But when trying to backup to BackblazeS3 I haven't yet found a configuration that works.
I created a new key pair for the exercise, and which connects using MinIO/client (mc), so I think the authentication is good. Perhaps authentication needs another step?
What I see in debug mode is
bareos-sd (100): stored/dev.cc:715-348 open disk: mode=OPEN_READ_WRITE open(icecubebareostest01/Full-0271, 00000002, 0640)
bareos-sd (100): backends/droplet_device.cc:357-348 check_path: path=<bareos-test/> (device="bbb" (icecubebareostest01), bucket=icecubebareostest01): Result DPL_FAILURE
bareos-sd (100): backends/droplet_device.cc:357-348 Retry: check_path: path=<bareos-test/> (device="bbb" (icecubebareostest01), bucket=icecubebareostest01): Result DPL_FAILURE
The bareos version is
20.0.0~pre1743.77a9856c8-8.el7
The droplet device is on a different VM from the director VM.
Thank you for any advice you may have.
James Bellinger