"test in a box"

"test in a box"

[David Gibson]

[-- Attachment #1: Type: text/plain, Size: 1050 bytes --]

Hi Jon (and others),

Are you still having trouble getting the passt tests to work?  I
mentioned at some point in the past that I had an experimental "test
in a box" script which set up some namespaces to run the tests
independently of the host's networking setup.  I just rebased that
onto the current tree and discovered that it worked better than I
recalled.  I was able to run the full test suite except for the
performance tests - I didn't try those, so I don't know if they work.

I pushed the rebased version, along with some other local patches I
tend to use while developing at:
	https://gitlab.com/dgibson/passt/-/tree/testbase?ref_type=heads

To use:

1) The test "assets" must be built and up to date, first, so in
   passt/tests run:
   	$ make
2) In passt/tests run
	$ ./testinabox.sh

Hope that's useful to someone.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: "test in a box"

[David Gibson]

[-- Attachment #1: Type: text/plain, Size: 1454 bytes --]

On Fri, Mar 15, 2024 at 01:42:17PM +1000, David Gibson wrote:
> Hi Jon (and others),
> 
> Are you still having trouble getting the passt tests to work?  I
> mentioned at some point in the past that I had an experimental "test
> in a box" script which set up some namespaces to run the tests
> independently of the host's networking setup.  I just rebased that
> onto the current tree and discovered that it worked better than I
> recalled.  I was able to run the full test suite except for the
> performance tests - I didn't try those, so I don't know if they work.

Sorry, now realised that's not quite correct.  It works for the subset
of tests that I usually use while actively developing.  That excludes
the log to file and memory tests, as well as the performance tests.
In any case, the change excluding those tests is also the branch below.


> I pushed the rebased version, along with some other local patches I
> tend to use while developing at:
> 	https://gitlab.com/dgibson/passt/-/tree/testbase?ref_type=heads
> 
> To use:
> 
> 1) The test "assets" must be built and up to date, first, so in
>    passt/tests run:
>    	$ make
> 2) In passt/tests run
> 	$ ./testinabox.sh
> 
> Hope that's useful to someone.
> 



-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: "test in a box"

[Stefano Brivio]

On Fri, 15 Mar 2024 13:42:17 +1000
David Gibson <david@gibson.dropbear.id.au> wrote:

> Hi Jon (and others),
> 
> Are you still having trouble getting the passt tests to work?  I
> mentioned at some point in the past that I had an experimental "test
> in a box" script which set up some namespaces to run the tests
> independently of the host's networking setup.  I just rebased that
> onto the current tree and discovered that it worked better than I
> recalled.  I was able to run the full test suite except for the
> performance tests - I didn't try those, so I don't know if they work.
> 
> I pushed the rebased version, along with some other local patches I
> tend to use while developing at:
> 	https://gitlab.com/dgibson/passt/-/tree/testbase?ref_type=heads
> 
> To use:
> 
> 1) The test "assets" must be built and up to date, first, so in
>    passt/tests run:
>    	$ make
> 2) In passt/tests run
> 	$ ./testinabox.sh
> 
> Hope that's useful to someone.

Actually, it would be generally useful to have this "on" in the tests
by default, either as a wrapper or integrated in a different way.

I maintain Debian and Fedora packages but I can only run manual tests
on them, and other packagers have the same kind of issue. In particular,
the lack of an autopkgtest possibility for Debian turns a release soft
freeze into a hard freeze:
  https://release.debian.org/bookworm/freeze_policy.html#hard

I know, the plan is to eventually replace this test suite altogether,
but your script is small and rather inviting...

-- 
Stefano