[Swan-dev] On re-applying "pluto: warn if loaded connection ended up unoriented" et.al.

Andrew Cagney andrew.cagney at gmail.com
Tue Jan 23 06:47:06 EET 2024


On Mon, 22 Jan 2024 at 19:40, Paul Wouters <paul at nohats.ca> wrote:
>
> On Mon, 22 Jan 2024, Andrew Cagney wrote:
>
> > commit b575e15e80bcf0924bc96e3e7420092becedc42a
> > Author: Andrew Cagney <cagney at gnu.org>
> > Date:   Mon Jan 22 13:32:01 2024 -0500
> >
> >    Reapply "pluto: warn if loaded connection ended up unoriented" et.al.
> >
> >    Log failed orient() during connection load using RC_LOG and not
> >    RC_LOG_SERIOUS.
>
> So technically, this is wrong. There is a very likely a "serious" issue
> for the user to be told. RC_LOG isn't in the category of "the user must
> be told". It's informational. So I would like this to be changed back
> in the future once the related bugs have been addressed.
>
> Also, please use separate commits for code and test cases in the future.

Except this wasn't my mess.

I was dealing with a commit that, once it became clear was broken,
should have been quickly reverted, followed by an incomplete trickle
of test changes.  Grouping them together as a single revert thus:

commit 2af2e7f6ac3096846d125ca05bfe4716b7bb6dc6
Author: Andrew Cagney <cagney at gnu.org>
Date:   Mon Jan 22 09:37:09 2024 -0500

    Revert "pluto: warn if loaded connection ended up unoriented" et.al.

    Back out change causing:
       strange log line addition causes "westnet-eastnet-auto" #1:
authentication failed: no certificate matched RSASSA-PSS with SHA2_512
and '@east' #1563
    along with associated test updates.

made handling them a lot less painful.  Ditto reverting the revert.

As for how we should be doing things just look at the below where
there's one test and one code change _and_ the pair score a 100% clean
bill of health from a test run:

commit 993c9e779a1a888707a64cb198e653b0a6f0cb74
Author: Andrew Cagney <cagney at gnu.org>
Date:   Mon Jan 22 10:09:37 2024 -0500

    orient: disentangle log and debug hw-offload fail

    and log at level RC_LOG, not RC_LOG_SERIOUS; RC_LOG_SERIOUS
    causes whack to assume the worst and immediately detach
    (leaving a connection missing keys for instance).

    Fix:
      pluto: log when only nic-offload setting causes interface rejection
    See:
      fail orient() when interface doesn't support required nic-offload #1527
      strange log line addition causes "westnet-eastnet-auto" .1:
authentication failed: no certificate matched RSASSA-PSS with SHA2_512
and '@east' #1563

commit c991edb929a1ecaa6710bd398ea8eed1d70c40ef
Author: Andrew Cagney <cagney at gnu.org>
Date:   Mon Jan 22 10:07:41 2024 -0500

    testing: simplify/rename addconn nic-offload VS hw tests

    mark as good; ref:
      fail orient() when interface doesn't support required nic-offload #1527


More information about the Swan-dev mailing list