On Fri, 2025-02-14 at 00:46 +0100, Dumitru Ceara wrote:
> <snip>

> > 
> > For the tests in ovn-northd.at, I think we should:
> > - remove unneeded DB table dumps
> > - reindent ovn-nbctl commands
> > - fix comments
> > - use fetch_column/check_column/check_row_count instead of bare
> > AT_CHECKs.
> > 
> > I was trying to cover the review comments (and taking care of the
> > test
> > nits) here:
> > https://github.com/dceara/ovn/commit/0785375
> > 
> > If it looks OK to you, feel free to use it in v7.
> > That commit doesn't add a test for the distributed NAT scenario
> > though.
> > 
> 
> Actually, I couldn't help it and wrote the distributed NAT test and
> it
> uncovered a bug in my incremental commit above.  The distributed NAT
> test and the fix for the bug are available here:
> 
> https://github.com/dceara/ovn/commit/119d849
> 
> The actual dev branch is:
> https://github.com/dceara/ovn/commits/refs/heads/tmp-bgp-lb-nat-advertise-v6/
> 
> The last two commits there can be squashed in patch 4/4 if they look
> OK
> to you too.
> 
Haha, I wish I noticed your mail sooner, would've saved me some time
:D. Anyway, I wanted to do testing in system tests, and in the meantime
I discovered one more bug. See my additional commit on top of yours:
https://github.com/mkalcok/ovn/commit/47d0c4bd6a80c6beb39e1cd5be730c2af7afb74c

Branch:
https://github.com/mkalcok/ovn/tree/bgp-route-leak-with-sb-v11

While I was stumbling around trying to hit the right combination of
options for distributed nat (I always forget all the conditions), I
found out that we need to check for nat record's validity as well. In
the tests that I added, I'm adding 2nd DGW port to R2, which causes all
the previous rules to be invalid (yet they somehow gain the
is_distributed=true).

I'll squash all these and post next version.

Thanks for the help,
Martin.

> Regards,
> Dumitru
> 

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to