| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-05-21 | |||
| 10:20:51 | sean-k-mooney | :) | |
| 10:20:51 | bauzas | ok, so, back to the pps spec | |
| 10:20:51 | bauzas | ok, so, back to the pps spec | |
| 10:21:03 | sean-k-mooney | brb continue and ill catch up | |
| 10:21:03 | sean-k-mooney | brb continue and ill catch up | |
| 10:21:28 | bauzas | here, we want to guarantee the process rate of the internal ovs switch that's on... br-int ? | |
| 10:21:28 | bauzas | here, we want to guarantee the process rate of the internal ovs switch that's on... br-int ? | |
| 10:21:38 | gibi | * gibi free until the top of the hour, then need to jump to two consecutive calls | |
| 10:21:56 | bauzas | * bauzas had to skip his usual Friday gym session :p | |
| 10:25:29 | gibi | in theory I we can rename the OVS Agent RP to br-int. It would help with some mental modelling. But this can be done any time independently of my work. So I would like to push this work to be a burden of the person who proposes the multiple OVS per agent spec. You can hate me for it, but I don't believe we ever get there. So I spare some effort not to do someting that will not be needed. | |
| 10:25:29 | gibi | in theory I we can rename the OVS Agent RP to br-int. It would help with some mental modelling. But this can be done any time independently of my work. So I would like to push this work to be a burden of the person who proposes the multiple OVS per agent spec. You can hate me for it, but I don't believe we ever get there. So I spare some effort not to do someting that will not be needed. | |
| 10:26:10 | gibi | I will documet the renameing idea in the spec | |
| 10:26:10 | gibi | I will documet the renameing idea in the spec | |
| 10:27:49 | bauzas | gibi: honestly, I feel we can just leave as it is, after 1 hour of brainstorming | |
| 10:27:49 | bauzas | gibi: honestly, I feel we can just leave as it is, after 1 hour of brainstorming | |
| 10:28:21 | gibi | I feel the same. but I think we gained some insights during that hour | |
| 10:28:21 | gibi | I feel the same. but I think we gained some insights during that hour | |
| 10:28:29 | bauzas | the main difference between pps and bw features is that one cares about inter-host communication while the other cares about the agent itself | |
| 10:28:29 | bauzas | the main difference between pps and bw features is that one cares about inter-host communication while the other cares about the agent itself | |
| 10:29:08 | gibi | bauzas: this is coming from the fact that the bottlenecks are in different places regarding bw or pps | |
| 10:29:08 | gibi | bauzas: this is coming from the fact that the bottlenecks are in different places regarding bw or pps | |
| 10:29:19 | bauzas | yup, that's what I figured | |
| 10:29:19 | bauzas | yup, that's what I figured | |
| 10:29:29 | sean-k-mooney | bauzas: kind of yes. one is inter host and the other is any traffic that passes though ovs bot inter and intra host traffic | |
| 10:29:29 | sean-k-mooney | bauzas: kind of yes. one is inter host and the other is any traffic that passes though ovs bot inter and intra host traffic | |
| 10:29:32 | bauzas | ok, I'll reply to the spec | |
| 10:29:32 | bauzas | ok, I'll reply to the spec | |
| 10:29:40 | gibi | bauzas: thank you | |
| 10:29:40 | gibi | bauzas: thank you | |
| 10:29:41 | gibi | sean-k-mooney: thank you | |
| 10:29:42 | gibi | sean-k-mooney: thank you | |
| 10:30:00 | gibi | I appreciate your time spent on these topics | |
| 10:30:00 | gibi | I appreciate your time spent on these topics | |
| 10:30:05 | bauzas | by saying that for the good or bad, we've already created a beast called "the agent" | |
| 10:30:06 | bauzas | by saying that for the good or bad, we've already created a beast called "the agent" | |
| 10:30:23 | bauzas | which is actually something between br-int and other things | |
| 10:30:23 | bauzas | which is actually something between br-int and other things | |
| 10:30:48 | bauzas | by adding resources to this "beast", we're making it clear this is the *switch* | |
| 10:30:48 | bauzas | by adding resources to this "beast", we're making it clear this is the *switch* | |
| 10:30:54 | sean-k-mooney | yes like the compute node it currenlty modesl multiple things | |
| 10:30:54 | sean-k-mooney | yes like the compute node it currenlty modesl multiple things | |
| 10:31:28 | bauzas | if someone pops up with wanting to monitor the number of ports per agent, we could then say "mmmm, maybe create a parent RP" | |
| 10:31:28 | bauzas | if someone pops up with wanting to monitor the number of ports per agent, we could then say "mmmm, maybe create a parent RP" | |
| 10:31:39 | bauzas | but this is unfair to ask gibi to pay the price now | |
| 10:31:39 | bauzas | but this is unfair to ask gibi to pay the price now | |
| 10:31:54 | sean-k-mooney | or just add it to the beast | |
| 10:31:54 | sean-k-mooney | or just add it to the beast | |
| 10:32:01 | bauzas | we just internally defined this as an agent, but this is untrue | |
| 10:32:01 | bauzas | we just internally defined this as an agent, but this is untrue | |
| 10:32:39 | bauzas | sean-k-mooney: yup, another spec, another day of discussions | |
| 10:32:39 | bauzas | sean-k-mooney: yup, another spec, another day of discussions | |
| 10:32:43 | sean-k-mooney | we shoudl add in scare quotes "the beast" beside all mentions of the agent rp and just not say why | |
| 10:32:43 | sean-k-mooney | we shoudl add in scare quotes "the beast" beside all mentions of the agent rp and just not say why | |
| 10:32:47 | gibi | until the agent and the switch has 1:1 relationship you can attach resources to the RP for both reasons | |
| 10:32:47 | gibi | until the agent and the switch has 1:1 relationship you can attach resources to the RP for both reasons | |
| 10:33:11 | gibi | it is a beast | |
| 10:33:11 | gibi | it is a beast | |
| 10:33:22 | gibi | but we are taming it with our 1:1 assumption | |
| 10:33:22 | gibi | but we are taming it with our 1:1 assumption | |
| 10:33:50 | sean-k-mooney | yes im tryiing to avoid discussign the added complxity hardware offload adds | |
| 10:33:50 | sean-k-mooney | yes im tryiing to avoid discussign the added complxity hardware offload adds | |
| 10:34:06 | sean-k-mooney | we have adress that via the different resouce classes however | |
| 10:34:06 | sean-k-mooney | we have adress that via the different resouce classes however | |
| 10:34:32 | gibi | which might be a compromise but as far as I see it is an acceptable one | |
| 10:34:32 | gibi | which might be a compromise but as far as I see it is an acceptable one | |
| 10:34:38 | bauzas | technically we created a beast in nova too | |
| 10:34:38 | bauzas | technically we created a beast in nova too | |
| 10:34:47 | bauzas | it is called a "compute RP" | |
| 10:34:47 | bauzas | it is called a "compute RP" | |
| 10:34:56 | gibi | sure it modells mutliple thing | |
| 10:34:56 | gibi | sure it modells mutliple thing | |
| 10:35:01 | sean-k-mooney | yes but its a natural starting point | |
| 10:35:01 | sean-k-mooney | yes but its a natural starting point | |
| 10:35:04 | gibi | :) | |
| 10:35:14 | bauzas | ok, I don't wanna hold us so long | |
| 10:35:14 | bauzas | ok, I don't wanna hold us so long | |
| 10:35:15 | sean-k-mooney | untill you need to decompose it it all in one RP | |
| 10:35:15 | sean-k-mooney | untill you need to decompose it it all in one RP | |
| 10:35:34 | bauzas | I'll just comment the spec and mention this very worthy eavesdrop | |
| 10:35:34 | bauzas | I'll just comment the spec and mention this very worthy eavesdrop | |
| 10:36:06 | ignaziocassano | hello, seems patches for bug 1815989 are for python3. I am using centos 7 train :-( | |
| 10:36:08 | ignaziocassano | hello, seems patches for bug 1815989 are for python3. I am using centos 7 train :-( | |
| 10:36:08 | openstack | bug 1815989 in neutron ussuri "OVS drops RARP packets by QEMU upon live-migration causes up to 40s ping pause in Rocky" [Undecided,In progress] https://launchpad.net/bugs/1815989 - Assigned to Rodolfo Alonso (rodolfo-alonso-hernandez) | |
| 10:36:09 | openstack | bug 1815989 in neutron ussuri "OVS drops RARP packets by QEMU upon live-migration causes up to 40s ping pause in Rocky" [Undecided,In progress] https://launchpad.net/bugs/1815989 - Assigned to Rodolfo Alonso (rodolfo-alonso-hernandez) | |
| 10:36:14 | sean-k-mooney | gibi are you happy you have all the answer to your question to proceed with the next version | |
| 10:36:14 | sean-k-mooney | gibi are you happy you have all the answer to your question to proceed with the next version | |
| 10:36:38 | gibi | I have one more open thing | |
| 10:36:38 | gibi | I have one more open thing | |
| 10:36:43 | sean-k-mooney | ignaziocassano: posibly but they shoudl be simple to adapt | |
| 10:36:43 | sean-k-mooney | ignaziocassano: posibly but they shoudl be simple to adapt | |
| 10:36:48 | gibi | do we need a microversion bump in nova for it? | |
| 10:36:48 | gibi | do we need a microversion bump in nova for it? | |
| 10:36:52 | bauzas | ah | |
| 10:36:52 | bauzas | ah | |
| 10:37:00 | sean-k-mooney | gibi: i do not think so | |
| 10:37:00 | sean-k-mooney | gibi: i do not think so | |
| 10:37:10 | sean-k-mooney | gibi: what api has changed | |
| 10:37:10 | sean-k-mooney | gibi: what api has changed | |
| 10:37:21 | gibi | nova adapts to the neutron API change | |
| 10:37:21 | gibi | nova adapts to the neutron API change | |