Encountered a tar ball unpacked inside an existing unpacked tar ball

[portante]

Observed on our staging server the following redacted output from find /srv/pbench:

/srv/pbench/public_html/incoming/6b9f18a77c40
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/.iterations
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/1-default
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/1-default/reference-result
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/1-default/sample1
.
.
.
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/1-default/sample1/user-benchmark.cmd
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/iteration.lis
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/metadata.log
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/pbench-user-benchmark__2023.01.04T20.50.30
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/pbench-user-benchmark__2023.01.04T20.50.30/.iterations
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/pbench-user-benchmark__2023.01.04T20.50.30/1-default
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/pbench-user-benchmark__2023.01.04T20.50.30/1-default/reference-result
/srv/pbench/public_html/incoming/6b9f18a77c40/pbench-user-benchmark__2023.01.04T20.50.30/pbench-user-benchmark__2023.01.04T20.50.30/1-default/sample1

It looks like the entire tar ball was re-unpacked as sub-directory of itself.

[dbutenhof]

What's the history of this dataset? How do we get a controller of 6b9f18a77c40? IPv6 hex, maybe?

[portante]

I believe this was uploaded from a container just recently, hence the weird host name, where the image is quay.io/pbench/pbench-agent-all-centos-8:latest.

[portante]

FWIW, another uploaded image using the same container image looks correct.

[portante]

And inspecting the offending tar ball does not show up any kind of redundant contents.

[dbutenhof]

Wow; now that's weird. 😦