podman mount: yet another race: layer not known

[edsantiago]

Triggered in parallel system tests with precious little to go on. I've reproduced via:

Window 1:
# while :;do printf "FROM quay.io/libpod/testimage:20240123\nRUN date\n" | bin/podman build -t foo -;bin/podman rmi foo;done

Window 2:
# while :;do bin/podman mount --notruncate || break;done
...
Error: layer not known

Took almost an hour to fail on 1mt. main @ 7899358 (includes #23645)

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[edsantiago]

Seen twice today, in the wild (not just in my pet PR)

• fedora-40 : sys podman fedora-40 root host sqlite
  • PR WIP: system test parallelization: two-pass approach #23275
    • 08-19-2024 12:08 in [sys] |060| podman run --mount image
• rawhide : sys podman rawhide root host sqlite
  • PR System tests: set a default XDG_RUNTIME_DIR #24073
    • 09-25 15:20 in [sys] |200| podman pod create should fail when infra-name is already in use
  • PR System tests: work around common gating flake #24069
    • 09-25 10:23 in [sys] |200| podman pod create should fail when infra-name is already in use
  • PR WIP: system test parallelization: two-pass approach #23275
    • 09-19 18:39 in [sys] |700| podman play with image volume pull policies

x	x	x	x	x	x
sys(4)	podman(4)	rawhide(3)	root(4)	host(4)	sqlite(4)
		fedora-40(1)