Thanks for the link to the CLT discussion. From a cryptography audit perspective, the idea of dropping python-cryptography is unfortunate. I think we need to move in the other direction (see https://tracker.ceph.com/issues/51946) - Ken On Wed, Feb 21, 2024 at 11:58 AM Adam King <adking@redhat.com> wrote:

This was discussed in the CLT call. There's some notes on it in the 2024/02/07 section of https://pad.ceph.com/p/clt-weekly-minutes.

On Wed, Feb 21, 2024 at 10:35 AM Matthew Vernon <mvernon@wikimedia.org> wrote:

...

Hi,

I'm referring to https://tracker.ceph.com/issues/64213 and the older https://tracker.ceph.com/issues/63529 (which relates to the dashboard mgr module specifically). But basically, the use of pyo3 isn't safe in subinterpreters[0], and it introduced a check to error out on such use.

This is already causing problems in Debian bookworm (I think my chance that a version of pyo3 with the check got released there first - it looks like both Ubuntu 23.04 and now centos9 are likely also affected?), although the MR that takes the problematic code out of the dashboard may help somewhat (and I've opened an MR to backport it to reef - https://github.com/ceph/ceph/pull/55689 ).

I may have missed discussion on it, but is there a plan for addressing this set of issues, please? It doesn't look like multiple subinterpreter support in pyo3 is going to happen any time soon...

Regards,

Matthew [0] https://github.com/PyO3/pyo3/issues/576 _______________________________________________ Dev mailing list -- dev@ceph.io To unsubscribe send an email to dev-leave@ceph.io

_______________________________________________ Dev mailing list -- dev@ceph.io To unsubscribe send an email to dev-leave@ceph.io