[Maarten Litmaath] 15:06:19 You know, oh Dear Storm, excellent. [Maarten Litmaath] 15:06:42 So, you're there after all. [Thomas Dack - STFC UKRI] 15:06:46 Sorry, yes, my last meeting ever run slowly. I was just getting so. [Maarten Litmaath] 15:06:52 So I just, told to the folks, that I've modified what you had. [Maarten Litmaath] 15:07:00 Create the. [Thomas Dack - STFC UKRI] 15:07:00 Yeah. [Thomas Dack - STFC UKRI] 15:07:04 No worries. [Maarten Litmaath] 15:07:08 Basically the proposed items [Maarten Litmaath] 15:07:16 So I did not, send an email to the to the working group. it occurred to me, but okay, So, I did. [Thomas Dack - STFC UKRI] 15:07:17 Yeah. [Maarten Litmaath] 15:07:25 What I set out to do. 2 weeks ago. And that is to. [Thomas Dack - STFC UKRI] 15:07:29 Yeah. [Maarten Litmaath] 15:07:31 To, merge all the easy, pull requests. So, it was, it was almost 10, I think, also, a bunch of them basically got superseded by by others so so they were all close to at the same time with a few issues that that were that were marked as resolved by these pull requests. [Maarten Litmaath] 15:08:05 What we. Yeah. [Thomas Dack - STFC UKRI] 15:08:06 And then that. That leaves us with 45 and 27 looking in it. [Maarten Litmaath] 15:08:13 45? No, I think should be a lot less, right? [Thomas Dack - STFC UKRI] 15:08:16 No, sorry, the numbers, for pull request, 45 and poor request, 27. [Maarten Litmaath] 15:08:20 Okay, yes, yes. And in fact, those 2, they, they are not difficult, but they, they require basically, so one of them is about the, meaning of the stage scope. [Maarten Litmaath] 15:08:37 And there has been a proposal. That we change slightly what it means. And that is basically the subject of one of those pull requests. [Maarten Litmaath] 15:08:52 Where basically is just a proposal to change the wording, but of course it has some implications on what people need to do. [Maarten Litmaath] 15:09:01 So essentially what we what we thought we would do was to have stage. As a superset of reed. [Maarten Litmaath] 15:09:09 Because the idea is why would you stage a file if not to read it? So, the idea was that that's if the file is already there. [Thomas Dack - STFC UKRI] 15:09:10 Yes. [Maarten Litmaath] 15:09:20 Well, you can immediately read it. But, Let's say the. Michael who who submitted that pool request felt let's say that this was sort of you know not compatible let's say with you know the spirit of tape systems and he sort of felt let's say that that let's say if you want to read that file then you should indicate explicitly that you that you want to do [Maarten Litmaath] 15:09:49 that and the stage a sort of a thing head that's can live on its own it can basically have its own timeline, which can be hours or days, even, if not weeks, potentially. [Maarten Litmaath] 15:10:02 So, so he sort of felt that say that we should not even suggest that we should not even suggest that a stage can very quickly suggest that a stage can very quickly be followed by an actual stage can very quickly be followed by an actual reed operation. [Maarten Litmaath] 15:10:15 And so he, so. I think this is fine. That proposal, but, and in fact, if you read carefully our version one profile, you will already see that the description of the stage scope is not consistent across the document because because in one place it is suggested that as a superset on the other hand the contrary is basically suggested. [Maarten Litmaath] 15:10:41 So, so already that has to be, has to be cleaned. Now, I could imagine, I say that in this, in this working group, we, we don't reach all the folks who need to pay attention to this, right? [Maarten Litmaath] 15:10:56 So, so if there's sort of no objection here I would basically draw the attention of the Doma BDT working group to this business to see basically where people feel. Yeah. [Maarten Litmaath] 15:11:08 Let's let's do that because basically nobody has implemented stage really and if they did They can easily make a change there. [Thomas Dack - STFC UKRI] 15:11:14 Yeah. [Maarten Litmaath] 15:11:20 Okay. So. So do some people already have an idea of well, let's not let's not try this or maybe the opposite. [Thomas Dack - STFC UKRI] 15:11:20 Okay, that's not sensible to me. What do others think? [Maarten Litmaath] 15:11:31 This is a very good idea. [Francesco Giacomini] 15:11:36 And so this is Francesco, so I can't really recall all the discussion. Yes, so I was trying to. [Francesco Giacomini] 15:11:45 To read it again quickly but I think we can discuss this among I think it affects mainly people with with this, with the tapes. [Maarten Litmaath] 15:11:57 Yes. [Francesco Giacomini] 15:11:59 And so probably we should discuss it, middleware, vendors say that have to do with tape, what do you think? [Francesco Giacomini] 15:12:07 So we, can agree among us. And then we we can provide feedback. Feedback here. [Maarten Litmaath] 15:12:12 Yes, or directly the ticket that would maybe be better. [Francesco Giacomini] 15:12:14 Oh, yes, yes, we have a in the working group or in the pickup, yes. [Maarten Litmaath] 15:12:18 I wanna take, sorry, the pool request or maybe there's open issue, but certainly the pull request can be commented on. [Francesco Giacomini] 15:12:21 Yes, yes. [Francesco Giacomini] 15:12:25 Yes. [Maarten Litmaath] 15:12:25 And the people can sort of chime in say well okay we can make this work for product X. [Maarten Litmaath] 15:12:34 So this of course has implications. Also for show for example. Yeah, because [Francesco Giacomini] 15:12:38 Yes, yes, in fact, yes, and FDS, we can we can discuss that we have a meeting list for the for the TPI, probably using that. Mailing list. [Francesco Giacomini] 15:12:47 And the people in board. So many. We we can we can reuse or abuse that that least. [Maarten Litmaath] 15:12:49 Yes, sounds like a promising thing. Yeah. [Francesco Giacomini] 15:12:56 So I can send the mail there. After this after this meeting. [Maarten Litmaath] 15:13:00 That would be good. Thanks. Yeah, thanks, Francesco. I'm not on that list because normally I don't really have. [Maarten Litmaath] 15:13:08 Have to be a part of such discussions. But if you [Francesco Giacomini] 15:13:11 Yes, Yes, but then we, I mean, if we find an agreement among ourselves, then I think that will be more or less. [Francesco Giacomini] 15:13:20 What is needed to close this pool in a way or another. [Maarten Litmaath] 15:13:24 Okay, and the other one is also for you basically, Francesco. [Francesco Giacomini] 15:13:28 Yeah, it was on this on the yeah, just another comment here. So in fact, there was something more formal here that there was something more formal here that test to do with the in case changing the version of the of the profile. [Francesco Giacomini] 15:13:41 And that was in the discussion. I don't know I see a comment from, well actually from the very beginning almost. [Francesco Giacomini] 15:13:51 Not the very beginning but at some point, Paul Miller in particular raised the the the point that we may change the version of the profile and or the version. [Francesco Giacomini] 15:14:07 Indicated in the scope in the in the access token. [Maarten Litmaath] 15:14:11 Yeah, so this is, well, okay, that's unresolved, at the moment. [Francesco Giacomini] 15:14:13 We can. [Maarten Litmaath] 15:14:17 There so and. Okay, we should, we should, we will have to, hash this out, let's say in authorization meetings. [Francesco Giacomini] 15:14:24 Afterwards. [Maarten Litmaath] 15:14:29 I was going to come to that shortly. But I wanted to you let's say to at least with your team take a look at the other pull request, okay, number 5. [Maarten Litmaath] 15:14:42 What was it? 2745. The one opened by Brian. [Francesco Giacomini] 15:14:44 No. Yes Yes. [Maarten Litmaath] 15:14:47 Okay, And then I commented and there was no for the follow-up. Okay, so we So. [Francesco Giacomini] 15:14:58 Yes, I think I remember being Yes, there was some confusion, so trying to have an abbreviated. [Francesco Giacomini] 15:15:09 In information in the what was in the scope probably [Maarten Litmaath] 15:15:15 So what we try to do in our token business was to sort of clean up a mess that we have inherited from, 20 years of, middleware, SRM and such. [Francesco Giacomini] 15:15:16 Yes. [Maarten Litmaath] 15:15:30 Which in the end, we sort of had to abort, to some extent because there was too much resistance. [Maarten Litmaath] 15:15:38 In particular, within Atlas, Atlas sites. About potential instabilities that this could cause so better basically gave up and said, okay, at least for now, we'll just have to, to make do, with what we can make, work. [Maarten Litmaath] 15:15:56 However, there is still sort of also this idea that, that within a token you would like to have a Yeah, bit of a, what is it, virtual namespace, shall we say. [Maarten Litmaath] 15:16:10 Where you would not have to know about the particular prefixes. Oh, that's a storage elements used to actually, store files for different VOs in different places. [Maarten Litmaath] 15:16:27 Now, so now we have a bit of an inconsistent situation. And, Brian, also trying to address a, well, a lack of documentation basically came up with that pool request saying, well, I think it should work like this. [Maarten Litmaath] 15:16:45 And he basically gave a few examples. But then in the eco pointed out, look, this has some consequences for at least storm. [Maarten Litmaath] 15:16:55 The way it is used, by, some of the, experiments or maybe by, yeah, in need, by some of them. [Maarten Litmaath] 15:17:03 And And, and we have, well, I won't say stalemate, but we have basically an unresolved matter, right? [Maarten Litmaath] 15:17:09 So I sort of said, look, on the one hand, we, We would like to say to do better in in the namespace than than in the past, but if we can't at least what we write up in the profile should be consistent. [Maarten Litmaath] 15:17:22 So we we should We should basically, describe what, let's say the storage elements are doing actually, but how they interpret. [Maarten Litmaath] 15:17:34 Paths in tokens. Because apparently it is it is not fully resolved strangely enough. So, so there is something working obviously, and because everybody supports tokens. [Maarten Litmaath] 15:17:50 But maybe. Some folks had to do a bit of a heck, to, to make this work. [Maarten Litmaath] 15:17:57 And so that is sort of the other thing where we. Again, here with a bunch basically of storage. [Maarten Litmaath] 15:18:05 Experts would sort of have to decide, hey guys, whether we like it or not, this is the current situation. [Maarten Litmaath] 15:18:13 And so that is basically what I was fishing for in that in that pool request. So whilst at the same time proposing a potential way to sort of move forward. [Maarten Litmaath] 15:18:27 And for the particular issue that was raised by, by, by, by Enrico, but is not, is not unique to Storm, so also D cache at a few sites. [Maarten Litmaath] 15:18:37 Has this, has this business particularly because LHCB Sort of wanted to do things that way. [Maarten Litmaath] 15:18:47 And, okay, it's maybe not our sort of job here to sort of try and fix these kind of things, but at the same time. [Maarten Litmaath] 15:18:58 We, it is our job to describe. How tokens should be interpreted. And if, let's say there is a bit of a clash, between the distinction of disc and tape storage elements because basically the same paths are used or at least the same prefix as I used then Maybe you need what I suggested. [Maarten Litmaath] 15:19:23 I suggested 2 or 3 possible ways that say to to slowly get rid of this problem and sort of have a cleaner situation in the future. [Maarten Litmaath] 15:19:32 Hey, I, I did not see a way yet to untangle these 2 things. But it's possible that after a few comments from people, we sort of, you know, are able to, at least conclude. [Maarten Litmaath] 15:19:49 This is how things should work today. Everybody implements it. Everybody can confirm as. So Storm, Dcash, XD and so on. [Maarten Litmaath] 15:19:58 They can confirm yes. If you come with such a token, this is what will happen. And so, so it's amazing. [Maarten Litmaath] 15:20:06 But, but it looks like we, still have some, some, things that are a bit ambiguous. [Maarten Litmaath] 15:20:14 In certain areas and and And also that, of course, has to be sorted in the next version of our profile. [Maarten Litmaath] 15:20:23 And so, so it's that not only let's say homework for you, your team, Francesco. [Maarten Litmaath] 15:20:29 But at least I'd say you guys could comment on, how you see the storm perspective. [Francesco Giacomini] 15:20:32 Yes, yes. Yes, yes, okay. Yes, we'll have a look. [Maarten Litmaath] 15:20:39 Okay. By the way, if somebody wants to say something, raise a hand, otherwise I'll just continue with the issue that Francesco already, mentioned had the fact that we that we need to decide. [Francesco Giacomini] 15:20:40 Okay, okay. [Maarten Litmaath] 15:20:55 Not now, but, but yeah, in the very near future. What to do with the versioning. [Maarten Litmaath] 15:21:04 So there has already been some discussion. I think there's also a at least one issue open by Paul, and it has Francesco said, and I think we've also discussed on the mailing list or so or maybe even a few of us between each other. [Maarten Litmaath] 15:21:20 So, on, on the one hand, of course, we have to be practical. And know that change on the production system is really slow. [Maarten Litmaath] 15:21:31 On the other hand, it's also not nice to sort of suggest here that we're still in version one at all. [Maarten Litmaath] 15:21:40 While making significant changes in the in in what we describe in the profile right so probably by now We have so many changes. [Maarten Litmaath] 15:21:51 That we can't really say we are still backward compatible with whatever we published in 2,019. [Maarten Litmaath] 15:21:59 And so. So we could, so that's a, the default is now that we would basically, decide at some point, hopefully this at the end of this spring. [Maarten Litmaath] 15:22:12 This is what we're going to call 2 dot o. It doesn't mean that all problems are solved, but at least I say we've cleaned up a whole lot. [Maarten Litmaath] 15:22:20 Clarified a bunch of things that were ambiguous or totally unspecified And, and then we also. [Maarten Litmaath] 15:22:27 Know where we, we need further work. That hopefully can then just be minor revisions, right? [Maarten Litmaath] 15:22:35 2 dot one, etc. And at the same time, we sort of felt that middleware should be a little bit more flexible in what kind of versions it accepts or not. [Maarten Litmaath] 15:22:50 And so we sort of, basically painted ourselves into a corner by having a very strong statement in the profile saying, well, you know, if If you, if you, If you, well, it's sort of like, well, what D caching sort of implemented in the end, right? [Maarten Litmaath] 15:23:10 If it isn't 1 at all. Then basically your token is refused. Yeah, whatever you put into your token, the version has to be bonded all because that's the only one that the cash people have, had some, some, some, some experience with. [Maarten Litmaath] 15:23:24 So we want to make that easier from now on. So Dave's already suggested. A few months ago, why don't we? [Maarten Litmaath] 15:23:33 Sort of say, you know, every minor release is not going to do any backward incompatible stuff. [Maarten Litmaath] 15:23:40 So if you accept 2 dot oh you should also accept 2 dot x. Anything that you don't understand you just ignore. [Maarten Litmaath] 15:23:53 Okay, so I probably say this is not controversial. Hey, but it means of course that we will have a big mismatch now between the, the published document, in a few months and what is on the production system. [Maarten Litmaath] 15:24:10 So I don't think it has to be the disaster. The the new version of the profile will simply indicate where we want to go to some extent. [Maarten Litmaath] 15:24:19 But of course, eventually, the cash and others will have to change cold. To, to be a little bit more intelligent about the diversion, the versions, that, are presented in tokens. [Maarten Litmaath] 15:24:33 Francesco. [Francesco Giacomini] 15:24:35 And yes, thank you. And so why in principle I agree that, well, with this semantic versioning of the profile and also the access So now I can't remember all the changes that have been applied to the to the document, the profile since 1 0. [Francesco Giacomini] 15:24:56 But, as you said, I think most of them were just the clarifications of something that was, unspecified or ambitious or a mistake. [Francesco Giacomini] 15:25:08 And so I would be reluctant to change, the, the version field in the, in the access token. [Francesco Giacomini] 15:25:19 Unless, of course I'm missing, some, feature that was introduced that was a backwards and compatible change or even a new feature that was not available before. [Francesco Giacomini] 15:25:30 Again, we can probably talk to the other vendors and say, if the changes that have been applied. [Francesco Giacomini] 15:25:40 Are really breaking the previous behavior that or say if there is a some client that can that can still apply the old behavior and saying I presented token with them. [Francesco Giacomini] 15:25:55 Version 1 0 and I pretend that that behavior is the satisfied. Instead of having the version 1 1 and so on. [Francesco Giacomini] 15:26:05 And So I don't know if there any modifications supply to the profile that, say can induce us to have, a change in the version of the token itself. [Maarten Litmaath] 15:26:21 Yes, so if you allow me quickly before, giving the mic to Dave to, to react to that. [Francesco Giacomini] 15:26:22 Yes, yes. [Maarten Litmaath] 15:26:27 I would defend the needs that we do not have a really really nasty change that say a really backward incompatible change in any of the changes that we have in our, in our profile. [Maarten Litmaath] 15:26:44 In practice. So the only sort of thing that we are now considering is to change the meaning of stage But basically it hasn't been implemented by anybody. [Maarten Litmaath] 15:26:53 So we can sort of say, okay. Nobody could depend on this functionality yet So we now, do it in a different way. [Maarten Litmaath] 15:27:01 But let's now listen to Dave. [Dave Dykstra] 15:27:07 Yeah, so because The protocol. Because the because the profile says that you have to, you know, reject anything that doesn't match. [Dave Dykstra] 15:27:22 We are now making an incompatible change to the profile to require. That the point versions are backward compatible. [Dave Dykstra] 15:27:32 Follow me there, so I'm saying that that. Profile even if the tokens not incompatible the profile itself is incompatible So that's why we need to make it a 2.0. [Maarten Litmaath] 15:27:45 Yeah. Yeah, but this is true, but I think they say that I certainly see where Francesco is going and I sort of hinted at that also myself that we We know that that production moves forward very slowly. [Maarten Litmaath] 15:28:04 So we can sort of pretend in production that we are still basically. Sending in it, version one that all tokens around. [Maarten Litmaath] 15:28:14 But we could need have a different, document versioning. And I think this was also proposed, I think, even by you, that we. [Maarten Litmaath] 15:28:25 That we could have essentially 2 version numbers. That of course ideally would match. And we could sort of put on onto our to do list to make the match again. [Maarten Litmaath] 15:28:38 But already say, look for Version 2 dot o we describe for example what you wrote so what you just said Dave this this backward incompatible behavior and we then put onto the to do list of the middleware project. [Maarten Litmaath] 15:28:54 That they should, you know, adjust the code to, to, to be able to do that, in the future. [Maarten Litmaath] 15:29:01 When at at some point, when we feel everybody is ready. We, we just, changed the version number in the tokens as well. [Maarten Litmaath] 15:29:09 And so, so I would sort of feel, let's say that the best we can do at this time is to sort of keep things more or less the same on production. [Maarten Litmaath] 15:29:20 Okay maybe we find a few places in some middleware where basically the behavior is not what we want or maybe, yeah, we could just label it wrong. [Maarten Litmaath] 15:29:31 So that wouldn't sort of be bugs to be fixed. But, indicate in our profile document where we want to go. [Maarten Litmaath] 15:29:39 And then, and then, and then. We will sort of discover her that yes, in, the, implementations that are there out there, the subset of all the stuff has been has been implemented. [Maarten Litmaath] 15:29:53 So not great. But maybe for the short term. We can sort of swallow this and then in the coming months, decide let's say how to how to move forward. [Maarten Litmaath] 15:30:07 So this is sort of my, yeah, my feeling it because if we If we just basically publish the version, the 2 that all document. [Maarten Litmaath] 15:30:17 Probably most of the middleware projects are gonna say, well, you know, it's going to take us some time to, to, to look into that. [Maarten Litmaath] 15:30:25 So, and what, what do we do in the meantime? And we would basically want to continue. [Maarten Litmaath] 15:30:31 Gaining experience with the tokens that probably in the end will not change very much. And so, so we would sort of not want to say, okay, have we now have version 2 that all. [Maarten Litmaath] 15:30:44 We need we need middleware providers to to react to that And then have all the sites deploy, you know, updates and or upgrades. [Maarten Litmaath] 15:30:56 So, that can easily be half a year or more. Okay, so. So that are the air so we should sort of repractical and say look we know where we want to go and namely to what 2 dot O, tells us to do. [Maarten Litmaath] 15:31:13 But we are sort of, you know, at an intermediary stage. Where we are, yeah, sitting between the original one. [Maarten Litmaath] 15:31:23 Dot O. Which already has not been implemented to the letter by anybody. And. [Maarten Litmaath] 15:31:33 In the middle between that one and the 2, 0 that hopefully we will publish in in a in 2 or 3 months. [Maarten Litmaath] 15:31:40 So, [Dave Dykstra] 15:31:41 Yeah, so some people could actually start to adopt some of the profile 2.0 things without change the token number. [Dave Dykstra] 15:31:49 Version number is what you're saying. Yeah. [Maarten Litmaath] 15:31:50 Yes, that's my whole. Yeah, that we can sort of, because, that's also why I need Francesco's question was so important that did we sort of break anything? [Maarten Litmaath] 15:32:00 Do we really want stuff to behave differently? Well, so far to my knowledge to say the answer in tokens not really so scopes are still still mean what they do with the potential exception of stage but there we we have a luxury that said had this other pull request, 45. [Maarten Litmaath] 15:32:22 We have to make sure that's say that we we come to sort of an agreement on what paths mean in in tokens because There, there is some room basically for a backward incompatible change, and then need we will have to See again if we can sort of, transition in a smooth way, forward. [Maarten Litmaath] 15:32:48 So I can give you an example, that already on the D cache site that there is, there's some flexibility. [Maarten Litmaath] 15:32:55 To sort of, so David, so Paul basically described an ad hoc algorithm already a year ago or so. [Maarten Litmaath] 15:33:03 That they can have a look at the path and if it doesn't seem to match Dave will basically say, [Maarten Litmaath] 15:33:09 That's maybe because the VO route was not included. So let's try again with the VO route. [Maarten Litmaath] 15:33:17 And so they already have, let's say, some flexibility there. And I was sort of hoping it that a similar trick could also be done on the storm side. [Maarten Litmaath] 15:33:25 If let's say there is a sort of a you know a block there again sort of that that we discover now that basically sort of you know makes me certain things impracticable impractical in in implementing and using Storm. [Maarten Litmaath] 15:33:46 And so, So I think this is certainly possible that we make such discoveries. Because after all in 2,019 We had 0 experience with this business. [Maarten Litmaath] 15:33:57 So we, We try to do our best for see all the situations, but but we probably missed a boat in in in a few on a few occasions and then now in 2,000 0 would be a good time to rectify that. [Maarten Litmaath] 15:34:15 Francesco, you still have your hand up? [Francesco Giacomini] 15:34:17 No, no, no, sorry, sorry. [Maarten Litmaath] 15:34:24 Okay, so again, Stefan, go ahead. [Stephan Lammel] 15:34:27 Yes, and I'm not sure since you mentioned this this guessing. If that is a good approach, right? [Stephan Lammel] 15:34:33 Because there might be a, it might be intentionally that user doesn't have, for instance, right access to the T pass and if his pass is not found on the disk endpoint, then automatically giving you access to the T pass might not be what we want. [Stephan Lammel] 15:34:53 So I think even that that guessing as in one has to consider that very carefully. So that I just wanted to point it out and regarding the, the version number, I'm seeing at the end is a search provide us, service providers need to support multiple profile versions, right? [Maarten Litmaath] 15:34:57 Yeah. [Stephan Lammel] 15:35:12 And I think, we cannot use some, 2.0 really until all those service providers have. [Stephan Lammel] 15:35:20 Provide the functionality for one and 2 simultaneously. And all the sites have upgraded. So every time we make such a change that will take a long Time to propagate sue. [Stephan Lammel] 15:35:35 But I think the storage for biters need to support both version one and 2 simultaneously so that we can switch because there's no way. [Stephan Lammel] 15:35:46 That on the experiment side we can. Keep track of which service at which side uses. [Stephan Lammel] 15:35:57 Which 4 5 version. [Maarten Litmaath] 15:35:58 Yeah, so this is another sort of practical problem, that we will, that probably we will need to look into, later this year. [Maarten Litmaath] 15:36:09 Basically, to sort of, yeah, see how we can do an intelligent rollout of of higher versions, because the bottom line is, we, cannot break production. [Maarten Litmaath] 15:36:20 So, so yes, you know, in some corners or so if sites were extremely slow to react, then, then then maybe that is that is permissible but essentially the damage should be very very very small. [Maarten Litmaath] 15:36:36 And so, so, and, and you, you said it, you, the progress will, will be slow and, and, we will have a need sites running different versions, and and we will have need sites running different versions of middleware that with different capabilities with respect to profile versions so so we we somehow have to find a practical way forward. [Maarten Litmaath] 15:37:01 But okay, this This, is a thing that, that we are not sort of trying to solve this year. [Maarten Litmaath] 15:37:08 So this is a thing that probably will evolve over the next years. But we hope, let's say, to, to have the baseline sorted out. [Maarten Litmaath] 15:37:17 I would still say by the end of this year. And, and then already, let's say our profile document should have caught up. [Maarten Litmaath] 15:37:26 And, and indicate. Hey, what's, where we want to go and, and so that people sort of know also, what they should still put on their, to do list for, for their, for their middleware. [Maarten Litmaath] 15:37:41 And yeah, we will keep discussing that in this kind of forum and and of course particularly the data management. [Maarten Litmaath] 15:37:47 Forums like the the DELMA BDT meeting at to see basically where we where we move forward without breaking production. [Stephan Lammel] 15:37:57 Maybe that we should already anticipate. Unsupported profile version and put that into the into the document as describing that if a particular the token comes with a profile that is not supported by the service. [Stephan Lammel] 15:38:16 Specify exactly what our message we want. Well, a rock port we want to get back. [Maarten Litmaath] 15:38:22 Yeah, yeah, that's another thing. Of course, that's 1 of your favorites, areas. [Maarten Litmaath] 15:38:30 Good error messages, but I think We all concur. [Stephan Lammel] 15:38:32 I also and also making the service providers realize that they have to provide that they cannot switch from one to 2 but have to provide multiple. [Stephan Lammel] 15:38:45 Bush and simultaneously. [Maarten Litmaath] 15:38:48 Yeah, okay, well, okay, here we probably can't solve it. Fully today. Well, what we do know and need is that needs to be sufficient so flexibility. [Maarten Litmaath] 15:39:00 So that in the end experiments can can keep going while while while at the same time we know that a steady improvement and we can do campaigns. [Maarten Litmaath] 15:39:11 And at some point we could even set hard deadlines if if it had to be. Hey, but hopefully, we, we, can be fairly flexible in the end, in, in these matters, but I don't think, let's say this is sort of fully crystallized yet. [Maarten Litmaath] 15:39:30 Dave. [Dave Dykstra] 15:39:31 Just wanted to make the suggestion that that you go through the open issues and put a tag on the ones that we want to get. [Dave Dykstra] 15:39:41 To be dealt with. Or 2 dot o like you could put a 2 dot ote Yeah, and the ones and the other ones are not that important. [Maarten Litmaath] 15:39:43 Yeah. [Dave Dykstra] 15:39:49 So. [Maarten Litmaath] 15:39:50 Yeah, that's a good one. I had anyway a plan to do something about this because, because indeed there are a bunch of issues that say that did not come with pool requests yet. [Maarten Litmaath] 15:40:00 So people, just pointed out, look, there is something not quite right here. And then, and then, and then, yes, there's a few in need that we should try and, and squeak, squeeze in. [Maarten Litmaath] 15:40:14 And, So one of my favorites, of course, is, is about all the token lifetime said that we have very sort of, strong recipes for that in the end are probably going to be unimplementable. [Maarten Litmaath] 15:40:29 To some extent. So, so also there we should at least indicate well in practice. This can shift a bit, but we can probably still indicate. [Maarten Litmaath] 15:40:40 A new sort of, you know, more realistic values for some of these things. Now that we have acquired quite a bit of experience with this this whole business. [Maarten Litmaath] 15:40:53 So, so also there let's say so I might even open another issue specifically about that and then of course in the coming weeks and months, see what we can do. [Maarten Litmaath] 15:41:03 So my hope is still, let's say, to, to have version 2 dot. Well, said by summer. [Maarten Litmaath] 15:41:11 Okay. If it gets delayed by 2 months or so, that is fine. Of course, but, but I sort of would like us to, to, to not have to keep. [Maarten Litmaath] 15:41:20 You know, waving. One dot O into everybody's faces and then immediately. [Maarten Litmaath] 15:41:30 Have to say, oh by the way, it's broken in a few places and you just have to ignore things, you know, where they are not okay. [Maarten Litmaath] 15:41:41 Yeah. That is not great. I think we can we can we can do better, right? It's now 4. [Maarten Litmaath] 15:41:46 No 5 years basically. Since we, road vandot of since we published one dot o. [Maarten Litmaath] 15:41:53 And so, It's about time to make something happen. I've put it onto my own, sort of to do list for this year. [Maarten Litmaath] 15:41:59 And, and by the time also Hannah is gonna be, back, fingers crossed and, and she may actually sort of, remember in particular how we made the beautiful PDF out of the, the GitHub machinery. [Maarten Litmaath] 15:42:17 So that it turns out to be surprisingly difficult problem. [Maarten Litmaath] 15:42:26 So if you allow me the other thing I wanted to point out, is that we, we need to device a list of urgent issues in I. [Maarten Litmaath] 15:42:41 Having to do with Let's say I call a VO management. Because that is basically what we are now facing that we have to say goodbye to vom. [Maarten Litmaath] 15:42:54 In 2 months or so. And, and then be fully reliant on I. And I've cooked up this Google Doc, linked from our, meeting, page. [Maarten Litmaath] 15:43:10 Bootstraps from, From Pete Volkage, a similar dog that he, is keeping track of for for Atlas. [Maarten Litmaath] 15:43:18 So it's quite funny, by the way, if you check out that document, you will see that, very large number of issues, have been fixed actually. [Maarten Litmaath] 15:43:29 Yeah, so, so he, kept them for the record, but he crossed all of them out. [Maarten Litmaath] 15:43:35 So that's quite a nice read actually to see that we actually made significant progress in the last year or 2. [Maarten Litmaath] 15:43:42 Yeah, but there is probably at least one and maybe 2 or 3 issues said that that I hope the I am developers can look into that that basically have some some consequences. [Maarten Litmaath] 15:44:00 Have for how a VO is managed. And, so so I hope a need head is, the Francesco, Can check it out. [Maarten Litmaath] 15:44:15 I've sort of, I basically anybody in common there. Add stuff or suggest let's say this is this is not urgent at all. [Maarten Litmaath] 15:44:26 We have a nice work around here or . So because we know that, well, there's, not a lot of time and it's not that people have nothing to do with so we should be careful with what we label as How about the one that for me stands out was was one opened by Patter, of course, about lacking emails. [Maarten Litmaath] 15:44:52 So email notifications about various, potentially bad things had that, VO members should, should be aware of. [Maarten Litmaath] 15:45:04 And Jessica. [Francesco Giacomini] 15:45:06 Yes, yes. First of all, thanks again for collecting to you and Peter before, for you to collect this this information. [Francesco Giacomini] 15:45:15 We have also transformed this into a project on Gitaba where we have collected in a dashboard. [Francesco Giacomini] 15:45:23 On the issues. I think we can put the May, the or, the chatter where we think we have included the open open issues that represent this what what we have written here. [Francesco Giacomini] 15:45:43 So, my, request is that, people, interested, look to see if we don't miss anything. [Francesco Giacomini] 15:45:52 And also it would be a very useful for people doing operations also for the experiments and that means for example that you use the dashboard. [Francesco Giacomini] 15:46:03 That can give some some some indication of what what the highest priorities are so that we don't just go ahead with our common sense which may be wrong. [Francesco Giacomini] 15:46:16 And then of course we will. We will do. Everything sooner or later, it's just that, we, want to be ready for, for the, for the transition. [Francesco Giacomini] 15:46:27 With the most important and urgent issue so if you can help us at least on this I mean you, you everybody. [Francesco Giacomini] 15:46:38 On this subject will be much appreciated. [Maarten Litmaath] 15:46:44 Yeah, okay, looks, I had a quick glance. So that is a very nice way to keep track of the stuff. [Francesco Giacomini] 15:46:44 And Yep. [Maarten Litmaath] 15:46:49 Let's say that is deemed urgent. As always, right, if you have a bunch of things, you think are are ready for for being tried out in production is so obviously just make a tag and and and then we'll follow the usual machinery here to try this out. [Maarten Litmaath] 15:47:10 On various instances. And, yeah, so then let's say we, with every upgrade, we can already improve from certain, uncertain fixes. [Maarten Litmaath] 15:47:27 But then again, A tag. Is still not completely cheap. So, so you, you folks have to decide, let's say when, when is a good time. [Maarten Litmaath] 15:47:41 Particularly of course yeah if there is something they say of deemed to a high priority and you think well we fixed it then then if we can try this out in production, then we can already profit from it. [Maarten Litmaath] 15:47:55 So, so we should we should just basically stay in in close touch. We have another, e- for that. [Maarten Litmaath] 15:48:03 So we're not going to bother everybody on the authorization working group pair with these these things all the time because it's really between the developers and and and people on the CERN side. [Maarten Litmaath] 15:48:14 And that's working very nicely. And so, yeah, I And by the way, you need it for those who didn't. [Maarten Litmaath] 15:48:24 We didn't know, but I think you need to, I may have said, is it, but we are sort of, you know, running the latest and the greatest. [Maarten Litmaath] 15:48:32 M. And Vomes in Porter. So So things are better now than they have ever been before. [Maarten Litmaath] 15:48:42 But we still have a few other kind of issues now to look into have there is a sort of, when we were gearing. [Maarten Litmaath] 15:48:53 Up towards the data challenge. There were, there were of course certain things that we were concerned about. [Maarten Litmaath] 15:48:59 But now, there's not going to be any, any high rate activity in the near future. [Maarten Litmaath] 15:49:06 So we can focus on a few other things and then, you know, look at the bigger picture again, as of, well, in the second half of this year, right? [Maarten Litmaath] 15:49:19 So, when basically we have this VOA transition behind us one way or another. [Maarten Litmaath] 15:49:26 So, okay. That's what I wanted to point out. I will sort of draw attention to to to those lists basically in a separate email I'll send probably later to the authorization working group. [Maarten Litmaath] 15:49:44 List, but I think that probably the folks like better. and of course Stefan who is in on the call. [Maarten Litmaath] 15:49:55 Are already in the and, Yeah, okay, so that is now sort of the the big thing staring is in the face right we we have this fairly hard deadline and and and we have a timeline that that that can work. [Maarten Litmaath] 15:50:17 By the way, Speaking of timeline. We also have. This other idea had to to move the NHC experiment I am instances from Open Shift to Kubernetes. [Maarten Litmaath] 15:50:33 That is also being worked on in parallel. But. We have Well, come to an agreement head that these things are, let's say independent. [Maarten Litmaath] 15:50:45 And so yes, we would like, let's say to start profiting from the Kubernetes deployment as soon as possible. [Maarten Litmaath] 15:50:52 The endpoints should soon become available so hopefully a matter of days to be tried out. [Maarten Litmaath] 15:51:04 In a first form and that would sort of be like like an imitation of what is there on Open Shift. [Maarten Litmaath] 15:51:11 And not basically well already offers advantages because the logging for example is better on Kubernetes and and one to other things but Not yet the full. [Maarten Litmaath] 15:51:25 High availability deployment that we that we have in mind so for that a bit of R&D is going to be done? [Maarten Litmaath] 15:51:34 It's already planned, but, we, we are not giving a, a firm timeline for that. [Maarten Litmaath] 15:51:41 So the whole push that say that we can. Sorted in the next few months. That's the expectation actually, but we we sort of said look at the very least there's going to be a second cluster that is talking to the same database as the open shift cluster. [Maarten Litmaath] 15:52:02 Can be used in parallel. Or Experiments can switch between one or the other, right? So of course the the second cluster the open the Kubernetes cluster will have to be tested, so there has to be a way. [Maarten Litmaath] 15:52:18 To point to that one. But the same cline should work. This has actually been verified by BAC, had the IM service manager at CERN. [Maarten Litmaath] 15:52:30 And he has even done a very high rate test. yesterday. So we has, we, he has set up 2 development, instances. [Maarten Litmaath] 15:52:40 Wanted to do the same database and he used the same client against both and both clusters to ask for tokens at a rate of more than 100 hertz. [Maarten Litmaath] 15:52:55 And, was able to, to run like that for 8 h, in, in a row with 0 errors. [Maarten Litmaath] 15:53:05 So, So that is very good news because it means that say that this this having shared so a shared database between 2 independent clusters does not lead to new race conditions very easily. [Maarten Litmaath] 15:53:27 Yeah, so, this was expected actually, yes. So, so it should behave like this, but okay, we said let's do a highway test and see you know if anything falls over. [Maarten Litmaath] 15:53:36 It didn't. And so, so that's very good news, I would say. [Maarten Litmaath] 15:53:42 And, yeah, okay. We will, when, these, the Kubernetes instances are going to be available for the experiments. [Maarten Litmaath] 15:53:54 That will certainly be be announced, well at the very least of course to the experiment representatives but maybe even as a generic email to the to authorization working group list so that everybody is in the know. [Maarten Litmaath] 15:54:15 And then if nobody else has something, there's 1 more thing maybe for the record. So I'm working for on a bit of training training, information for, VO administration. [Maarten Litmaath] 15:54:26 So, So, the idea is that, Probably it would be a large bunch of screen shots. [Maarten Litmaath] 15:54:36 In a single, presentation. That's, serve basically as documentation for, for what, what let's say a person in an experiment secretariat or soul might be confronted with and also what people trying to register in the in an LSE VO might see. [Maarten Litmaath] 15:54:57 And so that is my, my goal. And then of course this will be, be published in a convenient place and we can update it. [Maarten Litmaath] 15:55:07 As needed here. So this is, This is one of my highest priorities for the next week or 2. [Maarten Litmaath] 15:55:30 So does anybody, yeah? Go ahead, Tom. [Thomas Dack - STFC UKRI] 15:55:30 Catching up on loads. I was just going to. Chime in and add the. [Thomas Dack - STFC UKRI] 15:55:40 Between myself and Francesco at the team and we've been in conversations looking to organize another technical hackathon between the go I am as well. [Thomas Dack - STFC UKRI] 15:55:50 I need to do, send an email to the working group now that we have a date locked in, but we're looking to host this at the end of May, the 20 ninth and thirtieth at Knath in Bologna. [Thomas Dack - STFC UKRI] 15:56:03 Sorry, we'll send an email after this just with that information, but for those interested in participating in a technical hackathon activity. [Thomas Dack - STFC UKRI] 15:56:11 We will be hosting another one of them. In the coming month or so. Well, Francesco and the team will be hosting it, not us this time, but there will be one organized. [Maarten Litmaath] 15:56:16 Very good. [Thomas Dack - STFC UKRI] 15:56:22 Okay. [Francesco Giacomini] 15:56:25 Yes, you're all welcome, of course. So. In Bologna. It won't be hopefully too hot. [Francesco Giacomini] 15:56:33 The end of May, but who knows. Yeah. [Maarten Litmaath] 15:56:36 And we do have a way for remote participation. [Francesco Giacomini] 15:56:39 I think so. Yes, we can arrange something. [Maarten Litmaath] 15:56:40 Okay. [Maarten Litmaath] 15:56:45 Was that has worked out very nicely the previous times because Dana needs, yeah, you, were. [Maarten Litmaath] 15:56:54 You make it possible for more input to come at of course convenient times right so And so there was scheduling basically of slots and like, okay, this, this we will sort of, broadcast and be ready for remote input but of course most the hackathon will not be able to work like that. [Thomas Dack - STFC UKRI] 15:57:00 You do [Francesco Giacomini] 15:57:11 Okay. [Maarten Litmaath] 15:57:20 And so maybe for example the beginning of the day and and or the end of the day or so had to to have some some feedback options or so or input options. [Thomas Dack - STFC UKRI] 15:57:25 Okay. [Thomas Dack - STFC UKRI] 15:57:30 Yeah, something like that like we did last year where we can start with a develop date or something similar and then have an update on what people have been up to. [Thomas Dack - STFC UKRI] 15:57:40 Etc, and if we want to have some specific discussion sections or anything people want to talk about about. [Thomas Dack - STFC UKRI] 15:57:47 Wall. Like development conversations, technical conversations, and we can have that as well. [Maarten Litmaath] 15:57:56 Sounds very good. Thank you already. [Thomas Dack - STFC UKRI] 15:58:02 Does anyone else have anything they'd like to raise today? [Thomas Dack - STFC UKRI] 15:58:12 At which point I think for once we've actually keep it within the hour slot just so thank you for your time. [Thomas Dack - STFC UKRI] 15:58:20 I've got the notes ready here. I'll be uploaded. Awesome. [Thomas Dack - STFC UKRI] 15:58:28 Our next meeting is in 2 weeks time. If you, I will send an email for agenda requests ahead of time. [Thomas Dack - STFC UKRI] 15:58:37 So if anything, there, please let me know. Meantime, have a look and think around the. [Thomas Dack - STFC UKRI] 15:58:42 Open pull requests and the I am issues working doctors that we've talked about today as well [Thomas Dack - STFC UKRI] 15:58:51 Thank you very much, everyone. [Maarten Litmaath] 15:58:53 Thank you, Tom. [Francesco Giacomini] 15:58:56 Bye [Matthew Steven Doidge] 15:58:56 Yes [Maarten Litmaath] 15:58:57 Bye [] 15:58:58 Hi.