Hello, As a follow-up to this email, I did a patch introducing a per bridge-domain limit (https://gerrit.fd.io/r/c/vpp/+/30472). If it gets in, I was wondering if we should now deprecate the global limit (per option #2 below) or whether we should keep both (option #1 below). I am still under the impression #2 is better and I can work on a patch for that. Thoughts ? Jerome
De : Jerome Tollet <jtol...@cisco.com> Date : mercredi 16 décembre 2020 à 14:51 À : "vpp-dev@lists.fd.io" <vpp-dev@lists.fd.io> Objet : Bridge domains / mac learning limits evolutions Hello, With current implementation of mac learning in VPP, it is possible to configure maximum number of learned entries. This limit is global and shared by all bridge domains in a given VPP instance. I am considering implementing a per bridge domain limit making sure that a bridge domain can’t exhaust all available entries. I would like to gather the opinion from the community on that: 1) Shall we keep the global limit and add a per bridge domain limit? 2) Shall we only implement a per bridge domain limit and remove the global limit? My personal opinion is that #2 may be enough. It guarantees that the overall system can’t run low in memory because of too many learned addresses. However, it would change the existing behavior. Do people have an opinion about that ? Jerome
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#18407): https://lists.fd.io/g/vpp-dev/message/18407 Mute This Topic: https://lists.fd.io/mt/79000620/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
