# Transcript: 813464822
# URL: https://vimeo.com/813464822
# Duration: 4632s (77.2 min)

[0:02] So if no one is talking, feel free to just unmute yourself. Otherwise, just raise your hand and we'll go in order.
[0:10] And then, let's kick it off.
[0:12] So,
[0:13] Nico My name Yeah. Go ahead,
[0:17] Marilyn.
[0:18] Okay.
[0:19] What I'm asking this is a question I'm not I don't know if it's an aberration or if it happens all the time, but what I had was a a
[0:27] PGP key that expired,
[0:31] and I replaced it with a new updated key.
[0:34] But I I still didn't get the system to the routing to function right. I actually had to go down into the route and actually choose that key again in my routing.
[0:45] And I was wondering if that's if that is the way it works, because in special routing,
[0:51] I would think, man, this could cause me a lot of problems because if I change the key out, I could have, you know, who knows how many subscriptions
[0:59] that were running under that route. But I was just curious if you have ever heard of that happening before where you actually had to go back and choose the key again in your routing.
[1:10] So you're using advanced routing. Right?
[1:14] Correct.
[1:15] The key had exactly the same name like the old key? Exactly.
[1:20] It was identical.
[1:23] And when you selected the key, did you select it from the list or did you type the name? Well, you always type the name. It should have worked.
[1:32] So this doesn't make sense to me.
[1:35] Originally, I chose it from the list, the drop down. Once I gave it the part you know, the partner name, I was able to choose that key. There was only one available.
[1:43] But this would have been that routing wasn't in place, and months down the road, the key expired and had to be renewed
[1:51] by the partner. They sent in the new key. We imported the new key, but the route still didn't work, and we kept getting an encryption error. But we eventually got down to the point of, hey, let's choose that key again, then it then it function correctly.
[2:07] So
[2:08] this shouldn't have happened. So open a ticket with support and see what they'll say, because
[2:13] with the keys,
[2:15] it's based on the name of the alliance of it when you select it. That's what it looks like. Yeah. I mean,
[2:21] when it's identical, you would have thought to yourself, mother, there's no way that it that.
[2:26] But I appreciate the future.
[2:29] Yeah. I have a question on this too.
[2:31] Yeah. Just a second.
[2:33] Give me a second. So one thing, make sure you don't have a small big letter change in the name of the key. You know? Just cross your keys, make sure that it's exactly the same name. Because I have all got caught once by that.
[2:48] Annie, the way we do it, ours are always consistent.
[2:52] Always our partner name, uppercase,
[2:54] then we have an underscore,
[2:56] c u r r for current key, underscore, and p g p. It's always the same because
[3:03] I'd rather be consistent
[3:05] than have somebody say, hey, you're wrong. I've done it consistently wrong, and then fix that.
[3:11] But it's always the same.
[3:13] Okay.
[3:14] I fully
[3:16] understand,
[3:17] and that's the best way to do it. I was just putting it on the table. If you are absolutely sure it was the same one, check with support because this shouldn't be behaving that way as far as I know,
[3:29] simply because
[3:31] with the keys, we're looking them up by name.
[3:35] So
[3:37] if it's doing it differently, there might be so the only explanation is either that there is an ID hookup behind the scenes,
[3:46] maybe in the caches.
[3:48] So it might have been a cache problem, but check with support to you what's going on. Something something weird is happening. I heard someone saying they have another question on the same topic. So
[3:58] Well,
[3:59] yeah. Yeah, Annie. Actually, it's the same on the same issue here. Yep. Would it made a difference if
[4:07] you replace the key rather than delete it and then
[4:11] created a new one with the same name?
[4:14] Can you do that with PGP?
[4:16] Yeah. You can. I don't know.
[4:19] I as I said, in both scenarios,
[4:22] it should have worked the same way because
[4:24] the key on the the route only comes into play during the route execution itself.
[4:33] So I would expect that by replacing the key in any way or form, as long as the key is again there when the route triggers,
[4:42] that it should work.
[4:44] Yeah. It makes sense. But just maybe, for sure, when you create the ticket, explicitly say whether you deleted the p or replaced the key because there's a nuance there.
[4:54] Yep.
[4:56] Agree.
[4:58] And it can be
[4:59] as I said, it might end up being just a caching issue. If
[5:04] we are caching the key inside of that route,
[5:09] then it's possible
[5:11] that the change of the key on the account itself was not recognized
[5:16] for a while until you said, hey, the route changed.
[5:20] So this might need fixing. So I'm not sure if the problem was that the route was not seeing
[5:25] the
[5:27] new key as a new key or if it was not seeing a key at all.
[5:31] See where I'm going with that?
[5:35] So Yeah.
[5:36] Yeah. Just check with support, see what they might say,
[5:40] and let's let's go to the bottom of it. It shouldn't be behaving that way in any way or form. I agree with you. It it if this was going to behave like that, that's a lot of overhead on the admins to remember to go to save things. The whole point of hooking up by name
[5:56] was that as long as you name it properly, it will work. Right? Plus,
[6:02] especially
[6:04] because those keys can come from different accounts in different situations. Right?
[6:09] So check with them. My good feeling is that something got cached wrongly. Did you restart the servers in a before between
[6:17] the change of the key and the the resaving of the route?
[6:24] No.
[6:25] No. Yeah. I would have tried that as well. And if you have a dev environment, I will try that as well. If it is a cache issue,
[6:33] restart will will resolve it. At least you all know it's a cache issue.
[6:39] Thank you.
[6:40] Mhmm. Okay. So I have a couple of questions in the chat. How we monitor certificate expiration notifications
[6:47] in ST?
[6:49] I cannot pronounce your name. Vishnu
[6:53] Tas?
[6:54] How do you pronounce that name?
[6:56] Sorry.
[6:58] For anyone that has been on this calls before, I'm bad with names.
[7:02] So bear with me.
[7:04] Okay. Yeah. No.
[7:08] Can you hear oh, go ahead.
[7:11] So, actually,
[7:13] we we saw one application in the s three
[7:17] that talks about month the user expiration,
[7:21] locking,
[7:22] and disabling, and the PGP not notifications.
[7:26] Yep. So we wanted to use only
[7:29] we
[7:31] wanted to use this application only to monitor the PGP keys
[7:35] or other kind of keys.
[7:37] So how we can do it?
[7:39] Can it be explicitly used for certificates only?
[7:43] And if it is, then what kind of certificates it will monitor?
[7:47] So it will monitor all certificates
[7:50] inside of accounts.
[7:52] So if they're on the so let me open it. Give me a second here.
[7:59] Account maintenance.
[8:02] So
[8:13] see, here is the section.
[8:17] I
[8:18] you can just
[8:20] enable that one, specify how many days for the certificate.
[8:24] It will monitor anything that is under the account itself.
[8:28] If the certificate yeah. Is on this
[8:31] Go ahead.
[8:32] So it will monitor login certificates,
[8:35] BGP certificate, and the certificates
[8:38] But created
[8:41] for SSH servers and so on. Yeah.
[8:45] And how about the global certificate? Does it monitor those as well or no?
[8:50] No. This is only for the accounts.
[8:53] So if you are if you need to monitor the certificates over here on the setup menu,
[9:01] Those Mhmm. You need to monitor on your own. We do not have a mechanism for them.
[9:07] No. None.
[9:10] Okay.
[9:11] And another yeah. Question
[9:13] Go ahead. Yeah. But inside of the accounts, all three tabs, you know, if you go to an account
[9:20] and you check all of the three tabs hey. I don't have my account here yet. Yay. So over here, all of those,
[9:28] the login department, the private, all of them are monitored,
[9:32] but you cannot pick and choose. If you turn it on, it will be available for it will send you it will send a notification for all certificates here,
[9:41] anything expiring.
[9:45] Okay.
[9:47] Mhmm.
[9:48] And it can this application can be used explicitly for PGP keys, and we won't don't want to use it for accounts.
[9:56] Is it doable? Right? No. It well, can be used for all keys.
[10:02] But, yes, it can be used for just the keys.
[10:07] If you cannot make it to work, just put the account notifications to something like ten thousand days or stupid something stupid in the future like that.
[10:15] And then it will never work for accounts and then only for key. I think you should be able to set it apart for keys,
[10:23] but I've never done it personally.
[10:25] So in case it doesn't work, as I said, just, you know, make the account one somewhere in the far, far future and then the key on the short term and you're back in business.
[10:36] Yeah. This is one thing. Another thing I have is
[10:40] sometimes what happens is the same, PGP key certificate used under different accounts and, routes
[10:47] by making it public.
[10:49] So is there any best way to find out
[10:53] where all this particular certificate is used?
[10:59] So
[11:01] you know how I said about
[11:03] five
[11:04] minutes ago that,
[11:05] when you change a certificate because we reference by name,
[11:09] that's why it's so easy to find it. Yeah. You wouldn't adopt this scenario that because we're using it by
[11:16] name,
[11:17] if you're just using it by name, some kind of search might find it. But remember that we also have support for expression language over there. So it's possible that the key might be actually
[11:28] not even
[11:30] called by name,
[11:33] but by some kind of expression that extracted out from a variable, for example.
[11:38] So, unfortunately,
[11:40] the answer is nope. There is no easy way to find where it is used because
[11:45] we are referencing by name.
[11:47] If you don't use expression language in the PGP sections,
[11:51] you
[11:52] the easiest way might be to actually do an XML export
[11:55] and then look for the name of the the alias,
[11:59] you know, as old style as it it might sound.
[12:02] But
[12:04] if you're using expression language, for example, and something like flow parameters
[12:09] or account parameters to specify which key to be used for a specific account,
[12:15] then unfortunately,
[12:16] you know,
[12:17] you won't even be able to find it.
[12:23] And then in the yeah.
[12:26] And I have one more query.
[12:29] Sometime what happens is a partner we share the PGP keys to the partner, but they don't use that key to encrypt the files.
[12:37] And they use some different
[12:39] keys and send back to us.
[12:42] At that time, we go back to them and we ask what key ID that you you have used
[12:47] to encrypt the file. Then they share the key ID that we see
[12:52] in the certificate.
[12:54] Now we want to find out which certificate is with that key ID
[12:58] in our XVAE ST.
[13:00] So how we can find out that?
[13:05] Just a second for there was a quick question in the chat. The application we showed was called account maintenance,
[13:11] Jeff.
[13:13] I saw the question, and it's new in five five, newish.
[13:17] And he has a lot of nice thing, I guess, if you're not using it. So you have the ID,
[13:23] and you're trying to figure out
[13:26] which it might be?
[13:29] Correct. Yeah.
[13:31] That
[13:32] will be complicated.
[13:33] Essentially, you'll need to read all of the keys and
[13:37] with
[13:39] can you even do that?
[13:44] Am not sure if you can even do it. I I understand the use case. You're trying to chase down your key.
[13:50] I just don't know if we can do it with just the idea of the key. I've never needed to do that.
[14:03] So anyway,
[14:04] you can what your future implementation
[14:07] to get this
[14:09] in rest APIs also that we can have easy way.
[14:14] Yeah. I'll take a note. One thing you might want to tie
[14:18] is using the rest API,
[14:21] pull one of the keys
[14:24] that you know the idea of and see if the idea will show up somewhere on the parameters of the key itself. It should be somewhere.
[14:32] I don't know if you can search for it.
[14:34] So as long as it shows up in the IP itself, in the API,
[14:39] you
[14:40] should be able to just get the list of it, of all of them, and just look for it as ugly as it might be.
[14:47] But I'm not sure if we can search by ID.
[14:52] Okay. Yeah. I tried to I
[14:55] back, but I was not able to get that. That's where
[15:02] Okay.
[15:03] I took a note. I'll play with it a little bit more in the afternoon.
[15:07] If I find something, I will let you know. But I don't think I've
[15:12] never
[15:14] I know we couldn't in API one four. That part, can guarantee you.
[15:21] And in 02/2001,
[15:23] they did the new stuff. I don't think that this is one of the options for certificates key.
[15:29] But I'll try to check and at least to see if brute force might work. You know, I'll grab all the keys and then cycle through them.
[15:38] It's
[15:39] you know,
[15:41] even that I'm not sure
[15:43] how possible it will be.
[15:45] Also, keep in mind that
[15:48] as long as the
[15:51] the key is in a position where the account can see it
[15:56] for the decryption,
[15:58] it should be okay. But I it's indeed,
[16:02] it's very different. Right? During the encryption, you specify the key, you know, perfectly well which keys used during the decryption. The file just arrives.
[16:10] So usually, they encrypt with your keys.
[16:14] One trick might be to just put those keys out on the server tab, you know, on the certificates tab,
[16:21] and it'll connect the dots as you will will find them.
[16:25] You know, if you put your
[16:28] decryption keys on the certificate menu over here as opposed to inside of the accounts,
[16:35] the the route will pick them up from there because these are your global certificate.
[16:39] But I don't know how many decryption keys you have
[16:42] and how many you had given out. Right?
[16:45] If you're doing one on one, if you're generating a new pair for everyone, then that obviously doesn't work.
[16:51] So Got it.
[16:54] Okay.
[16:55] And one of the difficulties that we are facing is sometime what happens is the
[17:01] partner server gets changed
[17:03] from
[17:04] earlier name to some different name, maybe cloud migration or other activities.
[17:10] And at that time, we find difficulties
[17:13] in finding out which all routes that particular transfer site is being used.
[17:19] I tried to
[17:21] go to EP as well, but I was not able to get that. So And yeah. And it's Yeah.
[17:29] Go ahead. Sorry.
[17:32] So we wanted to know if there is any way easy way to get that information
[17:36] from the routes.
[17:38] Same problem.
[17:40] Remember then when you pick up a site and why I keep stop sharing?
[17:44] Because I you know what? I'll just keep the server on give me a second. Let me do some housekeeping here.
[17:51] So remember when you go on the route,
[17:54] inside of the step of the passing partner,
[17:58] you can either select the site, but you also can type an expression that will find the site on its own. Right?
[18:06] Yeah. That's where your problem begins. If it was again, if it was if you always name your sites directly
[18:14] inside of the route,
[18:16] you can I don't with the API, you can probably do style, but you'll need to circle to all of them? The easiest way is actually to XML export of that out and just look to them.
[18:28] But if you're using expressions,
[18:30] if you're putting kings
[18:31] like,
[18:33] you know, the account name or whatever,
[18:36] you will never find it that way because it will connect
[18:42] it will connect on on during the execution only. But if you are just looking to get that out, it will never connect.
[18:52] And that's where what is important is some naming conventions.
[18:56] But
[18:57] yeah.
[18:59] It it's the it's both the good and the bad thing in in the routing elements. Right? Because we're connecting by name,
[19:07] you can just build that out and just name things properly and they will work. The initial problem we started with notwithstanding.
[19:16] But it doesn't
[19:19] it doesn't lend itself to actually finding what it's used, especially with the expression.
[19:27] Access? Yeah. Use is fine. Use is just fine, but we are getting difficulties in finding in which all places
[19:35] that transfer site is used
[19:38] under the routes. Because we might have different routes,
[19:43] And since the account is made public, the transfer site will be available for any other accounts.
[19:50] Now making a change Yeah. Would involve an issue. Right? So
[19:55] in some cases, we might need to create a different transfer site
[19:59] and
[20:00] reattach
[20:01] under different routes.
[20:03] So in that case
[20:05] Mhmm.
[20:06] We need to find out which all routes are being used
[20:10] for that transfer site.
[20:12] So manually,
[20:13] if we have that's of routes, then it's very difficult.
[20:17] Yeah. Do you use expression language in the center partner selection of the site, or do you just select the site name?
[20:25] We just select the site.
[20:27] Okay. If you just select the site, then my best advice, it will be to use the XML export through the common line
[20:35] and just export the routes, not the,
[20:38] not accounts because they're with too much. You can export everything, of course. And then just search,
[20:44] and it will tell you where the the name is popping up. Of course, you can do that through the API as well. But because you need to go all the way down to the path to this to to the steps themselves,
[20:57] there won't be an easy way. It will be multiple steps.
[21:00] So with XML, it's the fastest. Plus, it will give you a number,
[21:04] and you will know when you didn't find all of them.
[21:07] And, yes, it's clunky.
[21:09] And, yes, we probably should care for better way. But, again, because on send to partner, you can use expression language,
[21:16] it makes it extremely hard to get you anything back out.
[21:20] It's in your case, you're in a better shape. If you never use expression language, at least you can search for them.
[21:32] It it's it sounds you know what? Go to the ideas portal and
[21:37] ask
[21:38] specify an idea saying that you want to look up
[21:42] screen where you can specify a site or a certificate, and it can tell you where s is configured to hard coded to use it. Let's say it like that. It will never catch the expression languages one, but it will catch your use case if they can build something like that.
[21:59] Okay.
[22:00] You know, that's the best I can tell. And in the meantime,
[22:03] don't forget to have the XML export. And I know the APIs are all the new and beautiful and so on, but
[22:11] old style always works,
[22:13] especially, you know, in your case.
[22:17] Makes sense?
[22:18] Yep.
[22:19] Okay.
[22:21] Back to the chat.
[22:24] Jort Jort, I don't know how to pronounce your name. Yes. Because you increment the yep. Jort. Right? Okay. Right.
[22:32] Because you increment the name of the certificate of the queue, it's a new I it's a new name, so you always need to go change it in that route. It doesn't it cannot
[22:41] figure it out. Right?
[22:43] Right. That's back to the what we're talking about, the PGP case.
[22:47] While,
[22:48] our first correspondent
[22:50] just replaced with the same name,
[22:53] in your case, you changed the name with the increment. And ST is not that intelligent. It doesn't know that it needs to look for the biggest number.
[23:02] So yeah. Right. That well, that's our policy to to change the name so we reflect the current year and the month and day on there. So
[23:12] you can see it.
[23:13] So it's it's a minor inconvenience, but it would be nice if we could just automatically replace it somehow.
[23:19] No.
[23:21] That would be the easy way.
[23:25] There's a there's a
[23:27] okay. So I went into the rest API
[23:31] for my ST production
[23:33] and looked up the certificates.
[23:35] And so
[23:38] you can actually pull it out with the API.
[23:42] But the ID,
[23:44] I'm not sure. I'm pulling it up by name,
[23:46] type, and issuer.
[23:48] You're saying that that it's not stored by ID or you can't look it up by ID?
[23:53] I don't know off the top of my head. I said that I'll play with it a little bit later. I know that there is no search mechanism for it.
[24:01] But if you do a get on the certificate and the certificate details,
[24:05] if the ID is there, obviously, it will store it somewhere. I think we should be we might be storing it somewhere.
[24:12] I just don't
[24:13] I don't think we can search by it and, you know, grab all 10,000 certificate to find one by a key by ID. It just doesn't sound like a great use of your time.
[24:25] No.
[24:26] So I do see the ID as a search in the query, and it does let you do it.
[24:32] Okay. I can't share my screen, but, you know It's okay. The screenshot in there? Okay. I put it in the screenshot.
[24:39] Yeah.
[24:40] Yeah. Okay. Yeah.
[24:42] Yeah. So so yeah. You can. You can actually filter by by
[24:46] the specific ID number.
[24:48] Yeah. See? One more problem solved. Question.
[24:51] Yeah. I had another question that I forgot.
[24:54] Yeah.
[24:55] Sorry. Thanks for looking it up, and then go with your question.
[25:00] Okay. Sure.
[25:02] You were speaking about the idea portal earlier.
[25:05] Mhmm. There's an idea that's gaining traction
[25:08] on there. It's s t dash I dash one forty.
[25:13] And it says no downtime when upgrading the s t cluster and edge. You probably have seen that. Right?
[25:21] And they were, yeah, we were talking about that in the in the the the group that we were talking about, the upcoming
[25:31] product.
[25:31] So there it's on the road map for but there's some confusion that it's not just for the cloud,
[25:38] but it's for actually SDI. And that's, you know, enterprise
[25:42] and maybe standards.
[25:44] So it's like that's my question. I use SD standard,
[25:48] and we don't really have any plans right now to upgrade to the cloud or to enterprise. So what are our options down the road, do you think?
[25:57] None.
[25:58] We will never ever have a zero downtime for standard course.
[26:03] Or let me not say never ever, but it's on the road map. No one is talking about it. The model itself doesn't allow it. So
[26:11] short of us rewriting the standard cluster in the near future, you're not going to get anything to help you with that.
[26:18] Sorry.
[26:19] So
[26:20] that so the only option would be upgrading to either
[26:25] enterprise?
[26:26] Would it be available for that?
[26:28] Or Yeah. Is it only gonna be available on cloud?
[26:31] So the
[26:33] first,
[26:34] the first will be for cloud only
[26:37] because, you know, we need
[26:39] a patient zero.
[26:42] The plan, as far as I know, is to try to make it for any enterprise closed up. How it will work and what it will require,
[26:50] they don't have the details yet.
[26:52] So standby on that. There will be more news as the year is as the year is rolling. But that's one of the main top big topics we had been talking about internally,
[27:03] zero downtimes and eliminating the downtimes on upgrades,
[27:07] especially with our patching now every month and so on. So
[27:11] standby
[27:12] is the best I can say. It's way too early. They are still working out the kinks.
[27:18] The part that we had been taught is that they're all been nothing for standard cluster,
[27:21] and it kinda makes sense with this replicating databases.
[27:25] You know? It just almost
[27:27] doesn't have a technological way.
[27:30] And I don't know yet, given you an enterprise cluster, will it be doable just with the cluster or if you will need additional software or hardware around it to support it.
[27:39] So Well, that might be a justification for us internally,
[27:43] you know, having a discussion about whether we wanna upgrade to at least enterprise
[27:47] if it's available for there or to cloud.
[27:50] So that's good. And, you know, enterprise cluster is not only just bigger than the standard cluster. It gives you a lot of other things, including
[27:59] the servers come back on their own without you needing to be up at 3AM in the morning to sync them in.
[28:04] So
[28:05] and, you know That's attractive.
[28:08] And this is already there. Right? This is how that's one of the main things. And
[28:14] so if you're looking for just the
[28:18] always
[28:19] these days, I think of the standard cluster almost as a standard model.
[28:24] It works, but the moment when your business is critical and you don't want to have someone twenty four seven on call for this cluster, going to enterprise actually makes a lot of sense. I mean, someone kicks the cable in your lap
[28:36] tomorrow
[28:37] and,
[28:38] you know,
[28:40] someone needs to be up to receive the clusters.
[28:44] Right. Yes. I I know exactly what you're talking about.
[28:48] So perhaps maybe this is a better
[28:51] topic for
[28:53] Nicole.
[28:54] But
[28:55] perhaps somebody should go out and update that idea and just kinda, like, set the record straight because there's a lot of you know, it's like I've got three notices today that that idea has been
[29:06] you know, had different questions and people are expressing confusion.
[29:11] I like the the platform change where it's actually
[29:14] you know, if you vote on an idea, it will, you know, give you all of the status updates whenever somebody adds a comment to it. But I think that that's something that, you know, customer service wise that that could be you know, set the record straight. Plus in the road map, it should be
[29:29] designated as like, okay. We're gonna do this for first for
[29:33] cloud and then for enterprise or whatever. So that kinda gets, you know, people, you know,
[29:40] focused on it to where there's less confusion. That's just
[29:44] an idea. Maybe I should put that in the idea portal.
[29:47] Part of it is that it's so new at the moment that some of the things might change by the time they kick it out of the door.
[29:54] You know, it's technology. If it turns out something is easier, they might keep up in a different direction. I think that's part of what's going on.
[30:03] The last I heard was cloud first and enterprise clusters, but,
[30:07] you know,
[30:09] ask me again next week that might have changed, and it's all my so for things like this.
[30:16] Of course. Wait five minutes and it'll be different.
[30:19] Absolutely.
[30:20] And because
[30:21] it's a road map,
[30:23] I cannot promise anything. You know, I'm implementation.
[30:27] And in the meantime, I'm I'm seeing that Joe Joe Joe, thanks for that, put the tank, and he made I was wrong. He was there in one four, and I forgotten about that.
[30:40] Mhmm. So
[30:42] You're welcome, Annie.
[30:44] Yeah. You know, I don't know everything. I forget things. That's why we have those meetings. It's all about the
[30:51] I I I I promise. I didn't set it up so that people can help. I really didn't remember.
[30:57] It's early in the morning here.
[31:00] Does breaking the cluster for a standard
[31:02] cluster kinda count a zero downtime? Is that recommended?
[31:06] Yes and no.
[31:08] It is a zero downtime in a way or at least lower downtime,
[31:13] but you're still down while you're breaking the cluster.
[31:16] And all the transfers that end up on one side of the cluster,
[31:21] you know, you can only do one way sync. So if you have two sides, two legs working at the same time, even for a minute with a single file,
[31:29] whatever goes through the one that will get overwritten,
[31:33] you will never be able to resubmit that file. The tracking table never gets merged again.
[31:38] We're trying that right now, so I'll let you know.
[31:41] Yep. And I've done done that a few times
[31:45] in the past, especially before enterprise cost driven existed.
[31:49] It works, but you just need to realize that
[31:53] you strap stuff. So if someone locks in so if you split into leg one, leg two,
[31:58] and leg one is your production and you start working on leg two,
[32:03] At one point, you'll need to resync either from, one to, like, two or vice versa. You know? That will override everything. So everything that happened onto the side that got overwritten
[32:14] will disappear. That mean login times, change passwords,
[32:17] any transfers that went through that,
[32:20] they they would have happened, you know, the files made it. But you don't have a record time anymore.
[32:27] Sure. Sure. Thanks.
[32:29] Okay.
[32:30] I
[32:32] see a couple of raised hand.
[32:35] Vishnu
[32:36] sorry.
[32:38] That's you again. Vishnu.
[32:39] Yeah.
[32:40] So right now, we are facing one more issue very frequently.
[32:46] One of the cluster node is doing offline,
[32:50] and the sync is not happening on these two.
[32:55] How
[32:56] this cluster?
[32:58] Are you standard or enterprise?
[33:00] Are you standard or enterprise?
[33:05] There's a bit of a Nico and some background noise. So if you're not speaking with Annie, can you make sure to unmute yourself,
[33:14] please?
[33:16] Otherwise, I'll do that for you.
[33:20] Okay. So sorry, Vishnu. Are you enterprise or standard?
[33:25] Enterprise cluster.
[33:27] Okay.
[33:28] So and there is no syncing in enterprise cluster.
[33:32] As soon as the server comes up, it connects to the same database,
[33:35] and it should be up and running.
[33:38] So
[33:38] No.
[33:40] What we have observed is when we let's say, when this one of the node is offline
[33:45] and we create first night or some changes in one of the node
[33:51] Mhmm. Ideally,
[33:52] our expectation is they should be replicated on the other node immediately.
[33:57] So sometimes either we are seeing lots of delay
[34:02] or sometimes they are not
[34:06] visible on the other node.
[34:08] That is one thing. And second thing is
[34:11] how what could be the reason on how to go to pinpoint the issue
[34:17] in such cases?
[34:19] Yeah. Your issues are caches, and that might explain also what you were talking about earlier.
[34:25] Basically,
[34:26] the two servers in an enterprise cluster talk to the same database.
[34:31] Literally,
[34:32] they go to the same place.
[34:34] However, each of them has a local cache.
[34:37] And if this local cache is not updated, it's possible that you have a big latency towards your database and the server is trying to compensate.
[34:47] It's
[34:48] possible that you have a misconfigured
[34:51] caches in your if you you
[34:53] haven't configured your caches properly,
[34:56] so the other server doesn't get the cache properly.
[34:59] So if this is happening,
[35:01] get support on the phone and get them to look to your configuration.
[35:07] Because
[35:08] you so enterprise clusters should be at instantaneous.
[35:13] Whatever you see on server one, you should be seeing on server two immediately.
[35:18] As in, I press the button, say, if you go to the other server, it's already there.
[35:22] Because in standard cluster, there are today separate databases. They need to be synced and fixed. Go wrong occasionally. Just ask people using standard cluster.
[35:31] With enterprise cluster,
[35:33] literally, it's a single database behind them.
[35:37] Right? Oh, yeah. So if you have a node that doesn't pick up immediately,
[35:44] get support to start looking at this configuration.
[35:47] You have something somewhere that is blocking the caching from building proper.
[35:52] Do we have any
[35:55] supporting document
[35:56] around this to check it?
[35:59] Or we need to raise directly a support case and ask them to
[36:03] provide a help on this?
[36:05] Get support to look into that. We have some so if you look at the capacity guide, there is some explanations of some configurations.
[36:13] But
[36:14] usually, ST on its own, if you didn't change the configuration of the caches, should not be doing what you're seeing. Something is happening in your environment. They need to figure out what's going on. It's possible that there is a someone changed the time out setting they they shouldn't have,
[36:31] or,
[36:32] there might be, as I said, latency to the database that might be causing issues,
[36:36] or it might be something in the kernel. I I don't even I haven't heard that kind of problem with enterprise cluster in years.
[36:45] Let's say it like that.
[36:46] It shouldn't be doing that.
[36:49] Mhmm. And what we just upgraded
[36:51] to
[36:53] the latest version of 5.5.
[36:55] And before there, it was working perfectly fine, and this issue was not observed at all.
[37:01] And this this has popped up after the upgrade. Okay.
[37:06] And this is important.
[37:07] Put that into your ticket and get support to look into that. This is a caching issue.
[37:14] Right? Oh, yeah. That part and especially
[37:17] if your PGP problem was happening around the same time,
[37:21] guess what? You it's the same problem.
[37:25] Okay. And another thing
[37:28] surrounding this,
[37:29] how
[37:30] the
[37:31] cluster status is monitored,
[37:33] like, in a when
[37:35] it says that
[37:37] one of the node is offline, who decides that,
[37:41] and how this happens internally?
[37:45] There is yeah. So there is an internal port
[37:48] between the cluster nodes,
[37:51] eight zero zero five,
[37:53] something like that.
[37:54] The the it is to the coherence model itself. You know, we have coherence on between
[38:00] so on ST itself, we have the coherence library.
[38:04] Just with Oracle, now it's the open one.
[38:07] What it does is basically pinging the servers. It's looking for case port.
[38:12] And if the port is not responding, if you have a latency problem, if it times out, it says, oh, I cannot find my friend anymore, so it switches it off.
[38:22] So it marks it as offline.
[38:24] If that is happening, you have networking issues.
[38:29] Yeah. And surprisingly,
[38:30] even after upgrade, it worked for a few months
[38:35] without any issues. Such kind of issues were not reported or but
[38:40] it just popped up since last three or four months, and there were no changes at the database or the network level that we were I'm sure.
[38:49] Yeah. Yeah. And that's you are aware, but check with your teams.
[38:53] It might be a DNS change. ST is very DNS prone to troubles. If the DNS changes around it and you don't restart the ST, ST can get a little grumpy.
[39:06] So
[39:07] check
[39:08] something changed.
[39:10] It might be our bill, but it might be something in the environment. But if you're having the server getting lost,
[39:16] and if you're having this severe cache issues,
[39:19] my gut feeling is you you have a latency problem or DNS problem in this environment somewhere.
[39:25] Also, and it's important, make sure the clocks are actually pointing to the same time on both servers.
[39:31] If they run out of sync time wise,
[39:35] you know,
[39:37] with if you if you're on Linux and you are not using NTP or something like that, you are not you
[39:43] are in a
[39:44] lot of trouble.
[39:46] But make sure that they don't
[39:48] if your clocks are not synced properly,
[39:50] the cluster gets a little sight wise because it really doesn't know what's going on.
[39:56] So that also
[39:58] might explain this kind of delays because,
[40:01] for example, you updated something at 9AM, but the other server thinks it's still eight for fifty nine. Guess what?
[40:08] Yeah. Got it. What you're trying. Makes sense?
[40:11] Yeah. So okay.
[40:13] And lastly,
[40:15] if you do this,
[40:17] assuming, let's say, we can land into situation where one of the node is on offline state.
[40:24] So how the transfers are still expected to happen from the both the nodes
[40:29] or only one? Well and that's where where the biggest problem is.
[40:34] Depending
[40:35] on why the other server think it's offline,
[40:39] it still might be possible that this server might still be processing, but it might not have the latest configuration.
[40:45] That's why you need to go to the bottom of it.
[40:48] Each server individually
[40:50] decides what jobs to run.
[40:53] So just because one of the cluster node says, I don't see my partner,
[40:57] does not necessarily mean that the partner is actually not properly connected and working.
[41:03] But this is the most dangerous time because it's possible that both of them are ranking running with different caches.
[41:11] And in this case,
[41:12] depending on where the transfer runs, it might behave differently.
[41:16] So get to the bottom of what you are saying. It sounds more and more like something in your environment
[41:22] is kept going a little higher.
[41:25] Okay.
[41:26] Thank you. Okay.
[41:29] Maryle,
[41:30] sorry about that.
[41:31] Your turn, and then I'll go back to the chat because I know I have questions there. So go ahead.
[41:38] What I was wondering is, is there anything on Horizon
[41:41] for
[41:42] Secure Transport to allow
[41:44] the
[41:45] exporting importing of routes.
[41:50] Of what? Oh, a routes.
[41:52] Routes. Yep.
[41:55] We have been asking. We don't have an answer yet.
[41:59] I not that I know of.
[42:02] Okay. I mean, I'm just I'm just a quick I mean, it would be nice to be able to you know, the thing that we talked the last time, Annie, about about this right here. But,
[42:11] you know, when you export import a partner,
[42:14] it would be it would really be great to be able to do one. You know, it's like it's like you figure out to yourself, hey. I'm gonna have 30 subscriptions
[42:22] for this partner. And guess what? Only 10 of them are going on into production right now. Right?
[42:28] I can't export the other 20 later on down the road and import them into production. I have to build them out manually
[42:34] because
[42:35] it doesn't allow for that. Or I have to drop everything I've got and then reimport it again.
[42:42] So I mean, I'm sure you've heard of that before.
[42:45] Yes. And your third option is to use the API,
[42:49] but it can be very clunky,
[42:51] if the route itself is inside of the account.
[42:54] You know, it's multiple steps. But as of right now, I haven't heard anything. This is the old conversation about promotion.
[43:02] How do we move partial account from one place to another, and how we reconcile partial accounts?
[43:08] Because if we're talking about the full account, export the account, import on the other some site,
[43:14] business done. Right?
[43:16] Yeah. Easy. Easy. But if you go down to the nitty gritty of I just want these five subscriptions together with their five keys and their three routes and
[43:26] a route template,
[43:28] that's not that easy.
[43:30] Simply because of the way the x XML import export works. And the API so so so you're kinda sorting the middle of it. So the a the with the XML, it's full accounts only.
[43:42] With the API,
[43:44] it's object by object. You cannot say, give me the, you know, all seven object. Now
[43:52] will we have a middle ground one day? We're hoping.
[43:57] But there haven't been anything yet on the road map.
[44:01] And I'll ask again. You know, I keep asking, but it's the promotion conversation.
[44:06] Unfortunately, we're nowhere near ready for it yet.
[44:10] Thank you. Okay.
[44:11] Okay.
[44:12] Back to the chat.
[44:14] I don't know who asked that, but so for air notification, is there a way for successful route notification to be filtered by direction?
[44:22] What do mean by direction? If it is a successful notification, it's only for pushes.
[44:27] The pulls are on a different place. So what's the question exactly?
[44:31] What am I missing? Hey,
[44:33] Annie. This is Morita from FTB.
[44:36] Hey.
[44:38] So we we have this route where we unzip the file, and we have to publish
[44:45] it back from the STFS going back to
[44:49] to the customer folder
[44:51] for it to be handled by
[44:54] by another simple route in there. And with that one, we noticed that
[45:01] the the customer
[45:03] received multiple notifications
[45:05] for all the for all the transfer
[45:08] that were flagged as route,
[45:11] as the, as route on the protocol,
[45:14] and this is for both inbound and outbound,
[45:17] transfer.
[45:18] So I was wondering if there's any way that
[45:20] are we if it's successful,
[45:23] I just want it to be sent if it's just an inbound
[45:27] it's just an inbound routing.
[45:30] Yeah. I know what you mean. You basically are republishing back in the same account.
[45:34] Yes. So, technically, your outbound becomes an inbound again.
[45:38] Yes. And it go to the same folder?
[45:42] Yes.
[45:44] Okay. One thing to check.
[45:46] Look at
[45:49] session dot protocol,
[45:51] I think, will be the variable.
[45:54] Look at the protocol.
[45:55] One of those
[45:58] things
[46:00] will have the protocol set as whatever they used for the upload, like HTTP or SSH. The other one will have the word routing on it. So this will be your differentiation.
[46:10] So on the condition of the route,
[46:12] I think it will be under session. Session dot protocol, if it's not there, look at the documentation,
[46:17] but find the protocol. I think that will be your easiest way and
[46:21] kick out the ones that are routing.
[46:24] That will mean that you only get notification on the first upload.
[46:29] With Makes sense?
[46:30] Yes.
[46:31] But
[46:32] in terms of setting it up,
[46:35] how do you apply that
[46:38] how do you apply that filter?
[46:42] Oh, when you When you create yeah. When you create the route over here.
[46:48] Come on, Wolf.
[46:50] Yeah. My server is sleeping.
[46:53] Now it's not. So when you so over here,
[46:56] where you have the route condition
[46:58] Mhmm.
[47:02] Session key. It's helping me.
[47:04] Like that. If session protocol
[47:07] and you probably want
[47:09] EQL,
[47:10] HTTP, or SSH, or router, whatever, and or not,
[47:14] probably do something like not session protocol. Yeah.
[47:19] Routing,
[47:20] and then it will not even go into the route. And if it's not into the route, it will not send notifications.
[47:26] Right? So you might need to build two separate routes. One, what will happen to sim what will happen with the file if it's the initial route or the initial upload, and one, what will happen with the routing one.
[47:38] Whatever you want to do. And then make sure you actually
[47:42] use the
[47:43] notifications on the routing itself, on the route and not on the package. Because if they're out on the package, any route will kick in, so that won't help you. But if you use notifications here as opposed to the one outside on the template on the package,
[47:57] know, you because you have the two spaces that that will do the take.
[48:03] So okay. So with this one, if it's
[48:07] so if I publish sorry. I'm I'm a bit lost on this one. So if I pub if I if I filtered what goes into this route based on the protocol,
[48:15] but I still have some steps that needs to be done with that. Right? And Yeah.
[48:22] You will need to have two separate routes. One of them running on
[48:27] the upload when it's not routing,
[48:29] that will do all your routing, whatever needs to happen and send to notifications if this is where you want them.
[48:35] And one running
[48:36] when the file arrives out from a routing that does other things but doesn't send notification. Oh, I see what you mean. I see what you mean. You know? Okay. Split the route. And I know that this will mean repeating the route twice.
[48:49] Mhmm. But at least it will give you the control.
[48:54] Yes. So, yeah, that's the cleanest way I can think of. Just split it by the protocol.
[48:59] And look at everything else under session. This gives you where the file came from. There might be other variables that will help you. But, basically, play with that. Play with the parameters, and that's the generic way to do it. Right? Anytime you need to find the condition. In this case, because the file is arriving from two different places, from different protocols, it basically, the easiest possible way.
[49:23] Okay.
[49:25] Okay?
[49:26] Alright. Thank you. Beautiful.
[49:28] One sorted out. What's next?
[49:30] Best
[49:31] practices for patching production Doctor environment,
[49:34] Michael.
[49:37] You're going with the hard questions today, aren't you?
[49:41] So
[49:43] are you enterprise cluster or are you standard cluster? Hey, Andy. Good morning. Yeah. We're enterprise cluster.
[49:49] Yay,
[49:51] Anthony.
[49:53] So
[49:53] the
[49:55] com the way to do it is step one, break the connection between two clusters, the database connection, so that the two databases are living alone.
[50:05] Now,
[50:06] are you on MSSQL,
[50:08] Oracle, or Postgre?
[50:11] We are neither we're on Oracle.
[50:14] Good.
[50:15] At least Oracle can be split because if your MSSQL
[50:18] is always on, that thing is a little weird. So
[50:22] the official line with production Doctor environment is that the step number one is go to your DBA to severe the connection between the two databases.
[50:31] At which point, you basically do have absolutely independent clusters.
[50:36] Right.
[50:37] Leave one of them in production so people can play with that and go upgrade the other one. And this is where something I said earlier about standard router router will come to bite you as well.
[50:49] Because
[50:50] at one point,
[50:51] when both both legs, you know, both the Doctor and the prop are already upgraded,
[50:58] you'll need to collect the databases again, to which point you decide which leg loses all the information and gets over it. So
[51:10] my recommend yeah. Go ahead. Well,
[51:13] so well, my question and, like like, I understand about breaking the database and stuff like that. I I I think, ultimately, my real
[51:20] question in this is that
[51:22] does that have to,
[51:25] does it have to remain broken until
[51:28] both prod and d r are upgraded to Yes. Its new version?
[51:33] Or Yes. Can the DBA create
[51:36] his
[51:38] snapshot standby,
[51:39] we upgrade d r,
[51:42] revert snaps
[51:43] shot standby
[51:44] after d r is down, revert revert it back to what it is, then go on and proceed with production.
[51:52] That way, it it's communicating
[51:54] well
[51:55] so my DBA is is is having the argument that it shouldn't matter
[52:00] if we're reverting
[52:02] the the Doctor
[52:04] database.
[52:08] Okay. And what He's not understanding why it has to stay broken, essentially,
[52:12] what I'm getting at. So so he he here is here is the easiest explanation,
[52:16] the easiest way to explain it.
[52:20] Because
[52:21] we're working with essentially
[52:23] exact replica of the database.
[52:26] As if
[52:27] you come if your production is on, say, the January release.
[52:33] Mhmm. And your Doctor is already on the February release. When you connect them, you will override database either
[52:42] and now the database will be either the January release database or the February release database. Mhmm.
[52:48] The other guys don't know what they're doing with that.
[52:53] That's your problem. That's why it needs to be split
[52:56] because the databases will be different. And how different?
[53:00] In some versions, it doesn't make a difference much.
[53:03] So they won't even care. In some cases,
[53:07] the the other thing might not even work.
[53:10] Okay. So, like yeah. Because because he was saying that,
[53:14] you know, we're not even updating the binaries on on the databases.
[53:17] So it it shouldn't matter. No. But we update the data and sometimes the structure of the data, but also some of the fields.
[53:26] So what might happen is that if you have the January 1 and working against the February, there might be a field missing or some values or a new index or something.
[53:36] Or if you have the February working against the the January working against the newer database,
[53:42] it might not know what to write somewhere. So you might corrupt an account by trying to save it because, you know, there is now a new field that is filled with it. So that's your problem.
[53:53] It's not
[53:54] the it's not the database itself. The database is okay.
[53:58] The problem is that the data insight
[54:01] is for the specific updates.
[54:06] And your biggest risk is actually data corruption.
[54:10] I wouldn't care if you just cannot see something or, you know
[54:14] but the problem is that if you're safe
[54:17] from the wrong site, it might either override in the wrong way. So now nothing is working, or it might not feel a specific
[54:25] field that now is mandatory
[54:27] because the binary is the best. They don't know about it.
[54:31] Makes sense? A report That would still be that would still be the case even though he reverted,
[54:36] the Doctor database back to its original
[54:39] sorry, previous date. I I I know that that's an extra step that that he's taking. So Okay. And I'm just
[54:47] And this is why I said I started explaining why we cannot wait. If he's reverting
[54:52] back so if you're so let's talk about numbers. So if if production is on build one and Doctor is on build two,
[55:00] and you rip and you should so you upgraded Doctor, you tested it, it looks okay.
[55:06] If he shuts down all the Doctor servers
[55:09] and revert
[55:10] the Doctor database
[55:12] to what it was still for build one. He can connect them at that point, but you cannot start Doctor at all until the database get swapped back.
[55:25] Oh.
[55:27] If
[55:28] you cannot start a server against the database
[55:32] that belongs to a different build.
[55:35] That's the important part.
[55:37] But even if it's replicating from prod?
[55:41] If if it's replicated from so if
[55:44] prod get upgraded in the meantime, it becomes on two, and it'll get replicated as two, and now it's again build two. Now you can start the DOS. Okay. Okay. So Just okay. So his his idea is correct. It's just I wasn't I wasn't fully understanding.
[55:59] Yeah. But I
[56:01] have a stupid question about that. I mean Okay. This will mean that you will be fully down until production is fully upgraded.
[56:10] It it the Doctor is down anyway.
[56:12] Okay.
[56:13] In this case not right. Yeah. That will work. In this case, that will work as a charm. Don't I mean
[56:19] And,
[56:20] quite honestly,
[56:23] you can even do some tricks that don't even require the database to go down, you know, but that will be a little bit more complicated and not as supported. So do it that way. But, yes, his idea will work. And that's why I wanted to explaining how it works.
[56:36] What you want to make sure is that if the server is if the SD is on version x,
[56:42] the database is on version the database it talks to when it starts is on version x as well. Other from that,
[56:50] how you achieve that, do you do a snapshot?
[56:53] Now one thing to be careful with the snapshot, although I don't think you have that problem because you won't be starting the year after you or a swap back.
[57:01] But
[57:04] and, you know, everything that you did in this upgraded database in Doctor will just get lost, but as you just use it for a quick upgrade, no one cares.
[57:13] So you are fine. Just but you will be down for for a long time.
[57:19] Usually, with Doctor, what I'm seeing is something little different, you know.
[57:24] Use the Doctor temporary as production
[57:27] until the the other server is getting worked on.
[57:30] Yeah. No. We we we do have that downtime scheduled. So it it it's already known.
[57:37] Okay. In this case, yeah, his idea will work as long as
[57:42] your OS team ensures that those upgraded servers don't try to go against this database.
[57:49] That's all you need to make sure Yeah. Which means, by the way, and I'll I I keep telling people that your first step during an upgrade will be to make sure that those servers never ever start automatically until you're done with the upgrading process.
[58:04] You know? Go and kick it out from the system or whatever you however you're doing your automate automatic. Because if your server get rebooted and this SD starts again to database,
[58:15] it's possible to corrupt things, because it will be the wrong database for a while. Yeah. Yeah. Yeah.
[58:21] That that's the only thing you need to be careful about. Other from that, Yeah. If he reverts back to your database,
[58:28] who cares? It's connected.
[58:30] Right. And then and then yeah. When we do prod, it would replicate to the newer version. Yeah. Okay. Okay. Perfect.
[58:38] Yes and no. Well well, yes and no. I mean, it would replicate from, you know, what is currently in production Yeah. To the ER. Yeah. Yeah. And here is where you need to be careful.
[58:51] It will not replicate
[58:54] database structure changes
[58:56] or anything that the upgrade process does itself.
[59:00] So here, we'll need to
[59:02] it will work, but the problem is that the in the he will actually sooner or later need to swap back with the temporary database that got upgraded on the Jira side anyway.
[59:14] Because,
[59:15] otherwise, it
[59:18] it won't
[59:20] the the the upgrade of the production will not fully upgrade the database on the other side.
[59:27] You still need to run through that as well.
[59:32] So
[59:33] I don't think that it will buy you as much as he thinks it will buy him.
[59:40] So the better practice on this is just to keep them broken until both environments.
[59:45] Okay.
[59:46] Yeah. That's the cleanest way.
[59:48] Especially if you never have users coming into the app. Because otherwise,
[59:53] you might
[59:54] so it will will update it, but the upgrader itself,
[59:59] let me think.
[60:01] What it will not do let
[60:04] me work in my head through that. So there are two types of information.
[60:12] The global one will get replicated.
[60:14] The locals are the same because
[60:16] your IDs are matching in VR. So in theory,
[60:20] that database
[60:22] should be okay on the other side.
[60:26] And here is where in theory goes. I don't know if it ever tested that way.
[60:32] Oh,
[60:34] that's the best I can say about that.
[60:37] Okay.
[60:38] Plus,
[60:39] there is another reason I like to keep them separate,
[60:42] and this is
[60:44] if something gets screwed up, you have a live database to swing from because
[60:50] the two databases are the same at this point.
[60:52] Mhmm. So if you update your upgrade your idea and it looks okay, now you have a database there that basically is the production database.
[61:01] Right? Because it was fully synced before you started.
[61:04] If something goes wrong during the production, if the database get corrupted, if someone does something stupid,
[61:10] you don't need the DBA to recover back from
[61:14] an older build, and you can actually work against the other database if needed. You can use the GR temporary as your
[61:21] pattern knowledge. If something looks weird, then the servers are upgraded. You can always link from the other database.
[61:28] I just
[61:29] I don't know. If you keep them connected during the process,
[61:34] you if something goes wrong, you basically screwed up every single entity you have. Yeah. No. They they they will not be connected. Like, he he's he was explaining that he was moving it from standby to a snapshot standby in order for us to upgrade the app And then reverting it back to standby
[61:51] after we were done.
[61:52] Yep. That's pretty much what he needs to do. Yeah. Yeah. That way, we would keep
[61:58] the production at the current level
[62:02] intact.
[62:04] Mhmm. But that yeah. Because, mean, we're doing it one day to the next. You know, we're doing Doctor, what, day one, and then we're doing prod day two. Like, I've already put this in test. It's been in test for three weeks. Oh.
[62:15] And, you know, we're we're we're okay.
[62:17] Yeah. Just keep them keep the connection between the two of them down during the night between the two days is what I'll say.
[62:25] Just keep them separate.
[62:26] The d r
[62:28] keep the Doctor down. It's database not replicating.
[62:32] Worst case scenario, you will lose whatever happens on the prediction between
[62:37] when you disconnected
[62:39] them and yeah.
[62:42] Between when you disconnected and when you go back to connected mode. That's your window. So for that one,
[62:49] what I would say before
[62:51] you go into the downtime for the production itself,
[62:56] do an export of the tracking table
[62:59] and everything else out
[63:01] before you start because it for the last day or so. So if you disconnected
[63:05] Saturday at lunchtime, for example,
[63:08] and you start to work on production on Sunday at lunchtime,
[63:12] export the tracking table and server logs
[63:15] between Saturday and Sunday because this is not on your replica database.
[63:21] Okay. So if something goes wrong, at least you have the the export. It won't be in the database. It won't be resubmitting all, but at least you have the records.
[63:31] Because if something goes wrong and you need to resync it the other direction,
[63:35] you'll lose all of that. Right. Yeah. It wouldn't work. Right. Okay. Yeah. So, you you know, just
[63:42] and it is just a precaution.
[63:45] But
[63:46] that happens,
[63:47] you know.
[63:49] Mhmm.
[63:50] It's software. Anytime you don't plan for something, it will happen. So I tend to tell people get as many
[63:57] backups as you can. And if you think you if you think you have a backup, just do another one just in case.
[64:03] That's the one time you don't, so you don't need it. Yeah. That's the time we'll need it. Yeah. And and that's our plan. You know, we're gonna do,
[64:10] you know, VM snapshots of the environment before
[64:14] we Yep. Apply update, and then right after we apply update. Yeah. And then we'll just complete those Yeah.
[64:20] When you do the snapshots, don't forget to make sure the test is completely down. Is down. Yes. Correct. As in, you know, PSEF,
[64:28] grab grab, make sure everyone is dead if you had go kill it.
[64:32] But make sure it's completely down so that you have a viable one. Because your enterprise cluster, you're actually in a better shape.
[64:39] Because even if you don't do that, there isn't much you can corrupt. You don't have a live database.
[64:45] But Right. That's the other thing. While you're doing those snapshots, get them to do the snapshot of the database at that point. They they are. They are. Uh-huh. Okay. They are. Yeah. Yeah. Good.
[64:55] And then do another backup just in case.
[64:59] The XML export.
[65:01] Just just in case everything shuts down, as you know, with an XML and system export, you can cover the environment very fast.
[65:07] So don't forget to do that as well before stopping the server.
[65:11] Okay.
[65:13] Yep. You know
[65:14] Okay. Good.
[65:16] Okay. Another question is, how can you stop,
[65:20] in progress transfer?
[65:23] Is this a server initiated,
[65:25] or is that a client initiated transfer?
[65:29] Hey, Annie. It is server initiated.
[65:34] Oh, dear.
[65:38] Not much you can do short of calling your firewall people and telling them to kill it.
[65:45] Oh, what if I told you that is
[65:49] not what's going on? It's it's
[65:52] it's not a network issue.
[65:54] No. I don't think it's a network. Yeah. I'm not saying it's a networking issue.
[65:59] What I'm saying is that
[66:01] with server initiated transfers,
[66:03] one day start, there is literally no way to stop them.
[66:08] Okay.
[66:10] It's a very odd problem and maybe a a situation that I should be addressing with support. But
[66:18] Okay. The system the system that the the
[66:21] system that ST is trying to pull the files from is actually
[66:26] up, and you can pull the files manually.
[66:29] But there are a large number of queue to transfers
[66:32] that are just seemingly stuck.
[66:36] That I might be here be able to help you. So if so
[66:41] okay. What version of STI on?
[66:44] 05/05,
[66:45] latest batch level.
[66:48] Which batch level?
[66:50] It's,
[66:51] 2023.
[66:53] Sorry. 01/26.
[66:56] What
[66:58] month is that? I don't remember numbers anymore.
[67:01] Oh, sorry. January
[67:03] 26.
[67:04] Oh, the January. Yeah. I might care about the kill for that. Let me get to the API page. We have a new API in the newer SKS,
[67:13] which is for the event queue.
[67:16] It's not on the UI. It's only on the API
[67:20] Oh. On the events.
[67:23] What you can get is do get here, and what it will get you will be what events are in the event queue at the moment. You know, the good old event queue where everything is stuck.
[67:36] If that event is actually in the event queue,
[67:40] you will see here
[67:41] that you literally
[67:44] can delete it.
[67:46] Oh, that is amazing.
[67:48] Yeah. We just added it a couple of versions ago. That's why I asked. So what it allows is to manipulate the event queue. So if you have 10 some idiot that put 10,000 files on his server without telling you, And now I still stuck trying to pull them. You can come to the API and actually delete those out from the event queue and clear your event queue. Very careful because delete is a delete. If you swap it, it never.
[68:14] Swipe it, it will never Now
[68:16] if
[68:17] because you're saying that the TM restart doesn't solve it, that tells me that those transfers are actually in the event queue.
[68:24] So
[68:25] look through the list with the event queue. It's not very easy to read sometimes, but who cares? You know, do your homework there. It's API only. There is no other way to get to that. It's API zero only.
[68:38] There is no,
[68:40] UI for that. There is no admin UI that I don't think they plan on one.
[68:45] But today, yeah, I get the list, find those, and just kill them, and that might clear it. Now why they're not transferring? That's a different conversation. You know? Chase it with support. But when they they get stuck that way, try to clear them first.
[69:00] Back in the days Okay. The only way for us to do that was to go out to the database, and we all know what that what
[69:06] disaster that might have caused. So that's the new way to do it.
[69:11] Okay. Thank you very much. That is so helpful.
[69:15] Yeah. And this this year now go ahead.
[69:19] Oh, I I was gonna say, isn't there a new global setting that you can
[69:23] do that to keep the the event queue clear for in progress transfers?
[69:31] There is. But
[69:34] you really have a normal
[69:35] program in progress. So from what it sounded like a disuse case, he just has stuck events that keep coming to point here at stats.
[69:44] The key is
[69:47] Go ahead. I thought the purpose of that that is to, when there's,
[69:51] events
[69:52] that get stuck for a long period of time, this, global setting will help it to time out, and it'll and SD will do some
[70:01] cleanup on that.
[70:03] It can, but it will require a start. It will require, actually,
[70:08] what what's up if you have a really long transfer that actually is going.
[70:12] So
[70:13] the event monitor will help you figure out what's going on first.
[70:17] Yeah. And that's that's the concern is that it's a very, very busy
[70:22] ST
[70:23] with a lot of big files going
[70:26] in and out. And,
[70:27] yeah, we're loathe to do anything that's going to impact anybody else who's using the system. So
[70:36] Oh, well, thank you. You know, I don't have that, that quite that issue, but I was we did enable that. But, Harsh,
[70:44] we don't have we don't send terabytes worth of files that were received in,
[70:48] usually.
[70:49] So Do you have the name of the the Joshua, do you have the name of the, property on your fingertips?
[70:59] That was You?
[71:01] George who was
[71:03] referring to the config option. Sorry. I don't. I'm not familiar. Yeah. I'm not familiar. Sorry about that. George, do you have the name of the property?
[71:14] Not off the top of my head. I know which one you're talking about. I just don't remember what it was.
[71:20] Yeah. It's a relatively new one. I think it was I I I
[71:25] yeah.
[71:27] Yeah. That's the January server, and I think it was already there in January.
[71:31] But, yes, it can help. But between so both of those changes, both this new property and the event monitor that I showed you, Both of them were added for the same reason.
[71:43] Because
[71:44] in the past, once something kicked into the event queue, it was basically there forever unless you touch the database, and we really don't like people doing that. Now between the two of them, you actually have a better control of what's going on.
[72:01] I think on the cluster, maybe.
[72:04] Okay. There's is it max retry account? Because I think the other issue that we have here, eventqueue.maxretry
[72:11] account, they're advanced
[72:13] routes.
[72:14] So
[72:15] there's, I know, some limitations on what can be done with those as well.
[72:20] I don't wanna do that conversation for too long. Yeah. That's a different conversation altogether.
[72:25] But also, don't forget that the event retry count,
[72:29] it has nothing to do with the file transfer retry
[72:32] count.
[72:33] It basically is the internal. So the event retry count is saying, if you're in if something goes wrong during the event, it can be any event, not just a transfer. It can be a delayed event or it can be a login event. If something goes wrong, how many times more
[72:49] to pull it out from the database before saying, you know what? I really can't do that. This is for cluster disruptions
[72:56] when the cluster node doesn't return it fast enough so that the appearance or the old model decides, you know what? This event is dead. Someone else need to catch it. It's when you restart the server, so the event goes into retry stuff like that. So it's it's different. It always did this property had always been there. It's our internal way to know when to to give up on things.
[73:18] Because we always try things.
[73:23] Thank you. Okay.
[73:26] Let me see. I don't have anything in the chat. We're almost on the half hour, but
[73:32] any last questions?
[73:34] Anyone else that have anything?
[73:36] And I don't think I have nope. No other questions.
[73:40] So anyone else?
[73:42] Five minutes.
[73:43] Last call for questions.
[73:52] No?
[73:53] Okay. Before I get it back to Nicole,
[73:56] if you are in the Phoenix area or the Scottsdale area, we will have in person user group in a couple of weeks.
[74:03] So come talk to me then.
[74:06] And I hope to see all of you on the summit.
[74:10] And
[74:12] okay.
[74:13] Last
[74:14] call.
[74:15] Questions?
[74:16] Anyone?
[74:19] Probably I should stop sharing.
[74:21] Thanks,
[74:22] Annie.
[74:24] Okay.
[74:26] That was a good session. I like it when people help. And, you know,
[74:31] when I don't know and I don't I don't remember, someone else does.
[74:34] So back to Nicole.
[74:37] Thank you, Annie, and thank you for this great session again. And thank you, all of you, for your time and discussions
[74:46] and participation.
[74:48] As Annie said, I hope that we will see you at summit.
[74:54] You will see that
[74:56] there is a cost
[74:58] for the summit, but it's a very reasonable cost that includes
[75:02] all the
[75:04] meals,
[75:05] the hotel for two nights in Europe,
[75:09] and for two or three nights for The US.
[75:13] The
[75:14] price for the summit in Europe is €300
[75:18] for the, gala,
[75:20] the
[75:21] entertainment.
[75:22] Of course, well, all the great presentation, there will be customers
[75:27] speaking.
[75:28] There will be use case presented by the PSO teams with customers sometimes. But,
[75:35] yeah, I'm very excited to have those
[75:39] summit again.
[75:42] So
[75:43] twenty three to twenty five
[75:46] in
[75:47] Europe at in Brussels
[75:49] and
[75:50] June
[75:53] in Arizona in Scottsdale.
[75:57] As I mentioned, you will be receiving
[75:59] an email that will lead to an article on the community where you will have the PDF of the slides that I've I've been sharing. So you'll have the link to register there.
[76:11] So, yeah,
[76:12] I hope we will be able to meet again.
[76:16] I will I won't join the one in
[76:21] Scottsdale, but I will be there in Brussels.
[76:25] So with that, I hope we will, see
[76:28] you, so either during summit or at least in, our next, user group. With that, I wish you all a good end of your day, and see you soon. Bye bye. Thanks, Nicole. Bye bye. Thanks, Nicole. Thanks, Annie. Thank you. Thanks, everyone. Have a wonderful day. Thanks, everyone. Bye bye. You. Thanks, everyone. Bye bye.
[76:52] Bye.
[76:56] Thank you.