ask-annie/out/838250027/transcript.txt

# Transcript: 838250027
# URL: https://vimeo.com/838250027
# Duration: 4638s (77.3 min)

[0:04] Okay.
[0:05] Who is going to ask the first question today?
[0:21] Hey, Annie. This is Ed Eccleston. How are you?
[0:24] Hey, Ed. Long time.
[0:27] Know. Long time no talk.
[0:29] Hope all is well.
[0:30] Yep. Okay.
[0:32] Go ahead. Hey. So so I I attended a an SD downtime
[0:39] council back in May
[0:42] Mhmm.
[0:43] Where where, I guess,
[0:45] actually, I was in the beginning stages of
[0:49] of designing,
[0:50] you know, how that how that would actually work with an enterprise cluster.
[0:56] I was curious
[0:57] if there's been any progress made on that front.
[1:00] Zero downtime is
[1:02] is a very hot topic here
[1:05] at Dell.
[1:07] And,
[1:09] you know, we have an enterprise cluster environment that we're,
[1:13] you know, we're wanting to
[1:15] you know, when we do patches or updates
[1:18] or
[1:20] things along those nature things along that those lines, right, we we wanna do so with
[1:27] with zero downtime if possible. So I was just curious
[1:31] where we are on the on on the Xplay side of things as it as it comes to that.
[1:37] Not yet.
[1:38] So they are working on it.
[1:40] We don't even have
[1:42] the reference
[1:43] architecture of how what you will need to do in order for it to work in your environment.
[1:49] But as
[1:51] long as there are any news,
[1:54] believe me, I probably will be opening this meeting with them.
[1:57] But I
[1:59] it's too early for that. They're in very early stages.
[2:03] As I said, we don't even have reference environments yet. We have our customers here in our cloud environment,
[2:09] obviously.
[2:11] We're not there yet,
[2:13] but that's the goal. The idea will be to allow you to do all the patching and requirements
[2:19] without actually losing anything and without checking down everything.
[2:24] What we know so far is that it won't be as easy as a rolling update,
[2:28] But that's pretty much all we had been told.
[2:31] So
[2:32] get me there.
[2:34] I'm just coming from summit. We had our summit last week,
[2:37] and they're a little further away compared to where they were back in May.
[2:42] But still, it's not fully built yet. They are still designing pieces of it. So
[2:47] it it won't be soon. If it is something that is blocking your updates this year, I wouldn't count on it. I don't have dates. Don't get me wrong. They might surprise all of us
[2:57] by eye, but I wouldn't think that we will have it this year yet, not for GA.
[3:03] So Okay. Okay.
[3:06] But it's coming. That's what
[3:09] we're all told.
[3:10] So Right. Right. Okay.
[3:14] And for anyone that that's listening,
[3:17] that will be enterprise cluster only. This will never show up for standard cluster for various reasons.
[3:22] So
[3:23] okay.
[3:25] And just just to follow-up on that
[3:28] Yeah. If you don't if you don't mind.
[3:32] So
[3:33] the other the other the other thing is
[3:37] is with the with the database itself. Right? So
[3:41] so I'm wondering
[3:43] if we're going to get to a point as well where we can do it, an automated database failover
[3:51] without interrupting
[3:53] the application.
[3:55] Right?
[3:56] So we use Oracle
[3:58] external
[3:59] the external database, we use Oracle.
[4:01] They have this fast start failover
[4:04] option.
[4:07] And So you know Mhmm.
[4:10] Yep. So I think you know where I'm getting at. Yeah.
[4:14] Not nothing on the map at the moment.
[4:17] However,
[4:19] it's possible that there's a side effect of what they are doing to that zero downtime. They we might be getting better at that.
[4:26] Okay. The current way it works is that we cannot tolerate losing the database.
[4:31] If whatever technology you are using, for example, Oracle RAC can swap databases behind the scenes without us even realizing it's happening on a single scan address,
[4:43] this is still working. It had always worked. It might be a little hinky sometimes as we all know if the Oracle doesn't read very clearly.
[4:51] But that kind of setup can't always work because for us, it's the same database, and we don't it it appears almost like just small timeouts here and there's we can't we what that we can't tolerate.
[5:02] But the way the software is written is that we use the database as configuration engine,
[5:09] especially the coherence layer on top of it.
[5:12] So the second we lose the database,
[5:14] we are kind of decking the water, so the demo start misbehaving,
[5:20] and I haven't seen
[5:22] anything. Now
[5:24] they are also working on something different, which will be the cloud edition of ST, which will be a very different ballgame
[5:33] that will be running in container, in clouds, and so on. And those will be able to tolerate things a little better, but that's a different conversation altogether.
[5:42] And this is even further down the road.
[5:46] Okay. So
[5:47] but,
[5:48] yeah, they are are working on things like that. We know that we overly rely on both the file system and on the database.
[5:57] But on the other hand,
[5:58] I mean, if we don't have access to the data, what can we do?
[6:03] It's not that easy to switch as many threads as many pulls as we have running all the time. So
[6:10] Mhmm.
[6:12] Plus Okay.
[6:13] There is a secondary the secondary concern here. When you're switching to a different database, how behind is that second database?
[6:21] Remember that we have the events table. We basically are controlled by the table in the database.
[6:26] And if you switch to a place where the event's table is even five seconds behind,
[6:31] you'll start presenting the same files and funny stuff like that. So that that's some of the challenges
[6:37] that they are looking into. But, yeah, I I know exactly why you're asking, and you need the full you
[6:45] need the perfect zero downtime. No matter what happens, you never go down. Right?
[6:50] We are working on it, but
[6:54] let's Yeah. We we we don't ever wanna be down, Annie. You know this one.
[6:59] You and everyone else on the phone, unfortunately,
[7:03] and I'll be very honest, no matter what we built and how we build it, outages will always happen. You know, you cannot predict everything.
[7:11] Sure. We're trying to minimize it. We're getting close.
[7:15] But how close? We shall see.
[7:18] So yeah. Hey, Annie. In in that session
[7:21] last last week, they talked about replicating
[7:25] the database
[7:26] and then pointing
[7:28] you know, swapping out in in a in a two node cluster
[7:32] if you will. You know, just pointing. And so you had you had one server
[7:37] down for patching while the other one was up and pointed at the replicated. It still
[7:44] was a little confusing in the sense that that five second delay you were talking about, that kinda you can drop a You'll transfer right then.
[7:53] You will always have that no matter what.
[7:56] Okay.
[7:57] It's
[7:58] we're getting too close. So because of the way the event table is written, because of how we handle it,
[8:04] dropping events will be a lot less likely than picking up the same one twice and controlling that in case your application is a little slow
[8:12] or a little old.
[8:14] So
[8:15] there will always be a point over there that you might be able to roll through it, but you'll need to to look into it. And that part of the challenges of what they are trying to solve.
[8:25] But it's an MFT system where depending on your system, you might be moving thousands of files per hour or
[8:32] more in some cases.
[8:34] You know?
[8:36] No disrupting will not be trivial.
[8:40] So we shall see.
[8:43] Okay.
[8:45] What do we have next?
[9:02] It's a very silent meeting.
[9:20] While everyone is thinking all questions,
[9:23] had you seen the new ST?
[9:25] I've seen the new UIs you are having.
[9:28] In case this looks your own unfamiliar, that's the main release of ST, actually.
[9:37] Oh.
[9:39] Hi.
[9:40] Annie?
[9:41] Yep. Good morning. I have a question related to the database.
[9:46] Mhmm. Mhmm. So
[9:48] do you suggest any
[9:49] specific database?
[9:51] Example, MSSQL
[9:53] or Oracle?
[9:57] Whatever your DBAs are more comfortable with.
[10:01] So I have DBAs
[10:03] for both.
[10:05] I mean Well,
[10:07] what OS are you earning on?
[10:12] I'm sorry? What OS sorry.
[10:14] Are your ST servers on Linux or on Windows?
[10:19] Linux.
[10:19] Linux.
[10:22] I am partial to Oracle simply because it's our oldest,
[10:26] and we have the most experience with working with it. And I personally have the most experience with Oracle.
[10:33] MS SQL,
[10:35] as long as you have a good admin team and that as long as they are not trying to cut corners on resources,
[10:42] can work as well. Postgre, our new kit on the block, let's say it like that,
[10:49] can also handle similar loads.
[10:52] So
[10:54] it really will come down to looking at what exactly you are doing.
[10:58] And unless you're trying to run something like a million files per hour or something, the database,
[11:04] as long as it's prop properly sized,
[11:07] will not really make much of a difference.
[11:10] So
[11:11] really comes down to what your environment is more comfortable with.
[11:16] Yeah. I now, we use
[11:19] Oracle large enterprise cluster
[11:21] on prem.
[11:23] Okay. But we are we are moving to cloud. So
[11:27] I think there is issue with the rack support.
[11:31] Mhmm. So so I was just thinking about
[11:34] because
[11:36] I mean, whether we can switch to the MSSQL.
[11:40] So that's the that's the reason I'm asking this question. So You can. The good news is that when you bring in the accounts, the XML export is database agnostic.
[11:51] Yeah. Yeah. So it doesn't care where you started originally,
[11:54] but we use both databases
[11:56] as,
[11:58] the same way, more or less. And even if under less schema is different, it's can be different.
[12:06] If we're talking about going to cloud, are you talk are you thinking about RDS or setting up your own database?
[12:17] Sorry.
[12:18] Which is what what is that first one? Sorry.
[12:21] Are you going to use the database as a service
[12:24] from the cloud team, or, is your team going to set the pre old databases in the cloud?
[12:31] Yeah. Our own database.
[12:33] Okay.
[12:35] Well, up to you. It's a depend I I cannot tell you what will run better in the cloud.
[12:42] We have both.
[12:45] So
[12:46] it's it's
[12:47] literally, if you're asking me Linux or Windows, I have a very strong preference. When we're talking about databases,
[12:54] it's just a preference just because I know Oracle better.
[12:58] But I also know it's very expensive,
[13:01] and sometimes, especially in the cloud, it's not getting situated properly.
[13:06] So,
[13:08] if you decide to switch, I would strongly recommend
[13:12] some
[13:13] performance testing
[13:16] before you go live because the parameters
[13:19] you will need to be adjusted, especially if you have high volume.
[13:23] Okay. It gets tuned a little differently from Oracle in terms of threats and so on because, you know, underlying technology is different.
[13:32] Okay. Good.
[13:34] The other thing is, as you know, or know, or don't know, haven't realized, we actually switched to open coherence.
[13:41] We're not using the Oracle coherence anymore,
[13:45] which
[13:46] kind of sort eliminated one of the reasons to go to Oracle for a lot of people.
[13:52] So
[13:53] but, again, if as I said, if you ask me, I always prefer Oracle.
[13:59] Again, pure personal preference.
[14:02] Our reference environment these days are with Postgre.
[14:06] So
[14:08] up to you. And we have quite a lot of customers on MSSQL by this time. It's not new.
[14:14] So if you need to switch to that data,
[14:17] it's okay. We have the information.
[14:20] So
[14:21] and I know you are looking for a yes, no answer and a little bit stronger worded one, but that's literally one of the cases where we can spend the whole conversation talking about what people prefer.
[14:34] And that's why my answer is whatever
[14:37] whoever is going to manage your databases feels more comfortable with.
[14:43] Make sense?
[14:44] Okay.
[14:46] Yeah. One thing, if you need to choose,
[14:49] I'll go with whatever you can get the bigger database in. If one of them will be a shared resource or if one of them will have restrictions on size or CPU, you know, I would go with the other one. Just general principle. You know? Give it more power, the better. That's that's the only thing I would look at
[15:08] if your admins are okay with both.
[15:11] Okay.
[15:13] Mhmm. Okay. Thank you.
[15:16] Of course.
[15:17] Okay. What else do we have?
[15:21] Hi. I need Joe here. Joe Hawk.
[15:24] Hey, Joe.
[15:25] Hi.
[15:27] On that zero downtime
[15:29] post, Greg, could we use that database instead of SQL Server or Oracle?
[15:36] It should work for all of the databases we have in enterprise cluster.
[15:41] Again,
[15:41] should bring the operative word for now.
[15:46] It's too early, Joe.
[15:48] I hear in. I hear you.
[15:51] But, again,
[15:52] our
[15:53] customer zero is our cloud, and our cloud had been switching to post gray in some environments.
[15:59] So I believe it will be for all of the databases, and that's what we had been told.
[16:05] The mechanisms behind the scenes might be a little different,
[16:08] but that's that's the story at the moment.
[16:12] Sure. Sure. So we'll see how that progresses.
[16:15] Right? Okay. Thank you. And if someone is interested in being an earlier doctor or early better tester or whatever,
[16:25] I believe they'll be opening the conversation for people sooner or later. They already are talking to some.
[16:31] It's
[16:33] we're not actually, we had changed the way we're doing things. Back in the days, we'll just build something and drop it and tell people go use it. These days, we ask a lot more questions early on.
[16:43] So
[16:45] that is that.
[16:47] Thank
[16:48] you. Okay.
[16:49] Mhmm.
[16:51] Okay.
[16:53] What is next?
[16:55] So so I have a quick question.
[16:58] Name is Carmen
[17:00] Burns.
[17:01] And
[17:03] I I'm kinda
[17:06] I've been using Axway, but not all the time. And so Mhmm. I I wonder if you if you create, like, a new account
[17:15] and you attach a git, a gid to it, but then Mhmm. The
[17:21] you you remove the gid and
[17:24] you put in another one. Is there any way that you could delete anything that's existing in the database
[17:30] for other accounts?
[17:35] Okay. Let's unpack that because I know where you're going. You're asking if we use the GID of someone else, can you get to someone else's files on the file system? That's what you are asking. Right?
[17:47] Well,
[17:47] I mean, what I'm asking you is is that
[17:50] there's you
[17:53] have all these accounts. Right? And I'm I'm
[17:56] I'm supposed to go in and create a new account. So I do. Mhmm. And
[18:03] so I added g a git an existing git, realized, oh, okay. That's the wrong one. So I remove it from this new account, and I get the right
[18:14] gid and
[18:15] add it. And so then in adding it,
[18:19] it's it's working right, but then the GID that I just removed gets deleted from the database and this associated with an account and it disappears.
[18:28] Is that possible or or what? I mean, I'm I've been Okay. Okay.
[18:34] Let me drop on the server because
[18:36] when you say GAD,
[18:38] you mean literally the value
[18:42] over here. Right? Hold on. Let me
[18:45] you mean
[18:47] this value here. Right? Absolutely.
[18:50] Yeah. So when you change it for a certain account,
[18:55] all it does if you're non root account
[18:59] if you are non root installed, if you're installed under a service account SD,
[19:03] it doesn't change absolutely anything
[19:06] except that some access rules might be tied to it and so on. But from the file system perspective, nothing changes.
[19:13] If you are root install,
[19:16] then what it will do when you change it, if it will it will ask you if you want to change the permissions on the file system. And if you say yes, it will change them and so on, but that's about it. It will not change anything for any other account at all.
[19:31] There is one exception.
[19:33] If your accounts are
[19:35] sharing their home folders
[19:37] or if they you have nested home folders where, for example, the supervisor is in user data,
[19:44] supervisor,
[19:45] and then everyone is under it. So,
[19:48] you you know, when you have nested folders.
[19:50] And if you change the current the GAD on the top folder, it will change for everyone.
[19:55] So they're under him.
[19:57] Oh. Besides
[19:59] that,
[20:01] whenever you change this UID, GAD here, it will only change things inside of the whole folder, which is listed on that.
[20:10] That's all. It doesn't change anything in the database. It doesn't change anything on the file system outside of this folder. And when you change it let me see if I can do that to show you actually.
[20:23] I don't remember if this server will give me that. But when you try to save it,
[20:31] yeah, my I'm non root. If I was root,
[20:35] it will actually ask me at this point, you're changing the UAD or GAD. Do you want to change the file system as well, and do you want to do it recursively?
[20:44] There is a a little,
[20:47] pop up now that will ask you what you want to do if you're on the newest versions, the last few versions.
[20:53] On the previous one, it was doing it behind the scenes for you.
[20:57] So on the updates before February maybe, it will just change it for you. So if you see changes in other accounts,
[21:06] then the only explanation for me might be that the other account
[21:11] actually had
[21:15] a nested folder. And there is one more option, and it might be just a user
[21:21] thing.
[21:23] ST is not TAP friendly.
[21:26] So if you open one account to look into it, and you then open a second TAP to open a second account,
[21:32] and you go to save the first one, it will actually save the second account with the data of the first one.
[21:38] That's eliminating one account. So
[21:41] maybe I don't know what happened to to bring your question, but you you be careful with that. They're actually
[21:48] as you as you saw the first page on the five five on the mailing list, they wrote the whole page. They're actually writing the whole application to make it a little bit more user friendly,
[21:58] but it takes a while. But I
[22:01] if you can tell me more about what happened to bring the question, we can go bottom to the bottom of what happened. But other from that, the answer is changing the UADGV of a single account
[22:13] does not change anything for other accounts
[22:17] unless they are nested under him.
[22:21] Okay.
[22:23] And, of course yeah. Go ahead. Sorry.
[22:27] I said that that could've that could've been.
[22:30] Mhmm.
[22:32] Yeah.
[22:32] I guess in going further, it it gets it gets comp well, not much more complicated than what you explained.
[22:40] Yeah. And yeah. UID,
[22:42] GID is not on the business unit level
[22:45] anywhere. It's just on the account level.
[22:48] So,
[22:48] again,
[22:50] if something changed elsewhere, it's one of those things. Either you had two browser windows open, so you save the wrong on top of it because it saves the last position regardless of where you are. Mhmm.
[23:00] Or,
[23:01] as I said, you had to invest it somewhere. So that one changed to another. If something else happened,
[23:07] it's a bug, but I cannot imagine how. Now if you had been changing home folders instead on someone else's folder, That's what happened there. Right?
[23:18] But, just with GAD, it shouldn't make a difference. Do you know if you are root or non root install?
[23:24] I'm sorry. Can you repeat that last part again? I missed it. Sorry.
[23:28] Do you use Root or non Root install?
[23:32] Is the application
[23:33] is Oh,
[23:36] it I mean, it it would be I would think non non Root, but, there's still a lot that I need to learn. I was gonna ask you for your best recommendation for someone
[23:47] kinda starting
[23:48] out. I mean, not completely starting out, but just I
[23:52] would like to know more about the
[23:55] about the application and, you know, the dive into, you know, creating accounts more often and
[24:02] understanding the different options.
[24:04] So Okay.
[24:06] So to answer that, look at what we have at the university,
[24:10] actually, in university site.
[24:12] There are some basic
[24:14] free classes
[24:16] that will lead you through some things and explain some things. There are also actually
[24:21] a real good advanced routing class.
[24:24] But, that's if you're doing going to do a lot of onboardings.
[24:29] But even with the free options, there is a few things that might be useful.
[24:35] Okay.
[24:36] And the other thing, we had increased
[24:40] a lot what we have okay. Stop sharing. Hold on.
[24:45] We had added quite a lot of,
[24:49] help pages all over the place. So and I know they are not that helpful for,
[24:55] the account itself, but when you start playing with subscriptions and routes, there is a lot more information above on the tooltip
[25:03] and in the help page.
[25:05] But if you are just starting again and you haven't done a lot of it, look through the free classes available
[25:13] from university
[25:14] from Maxwell University
[25:16] because they explain the basics of fields and stuff like that here and there.
[25:22] Okay. So you don't have one specific
[25:24] class in mind?
[25:27] You No. Not really.
[25:29] Yeah. Not really. Okay. Alright. Yeah.
[25:33] As I said, start with the free classes because we have a few that are just recordings that you can listen to at any time for free.
[25:41] And if you want more, obviously, we have the paid classes as well with a lot more detail.
[25:48] Okay. And a lot deeper.
[25:50] But at least
[25:51] to get you kick started, they actually made some things available
[25:56] initially,
[25:57] the basic stuff.
[25:58] Okay. Cool. Thank you. And from there, it will really depend on what you you will be doing on the server. Right? If you're going to be the onboarding person,
[26:08] you know, all the routing and so on,
[26:11] as I said, the advanced routing class is actually very good for that because it will lead you through every single option of
[26:19] all kinds of things of what you can and cannot do and how you can think about advanced routing
[26:25] because it it's a whole engine. It's a server inside of the server.
[26:29] Is it? Okay.
[26:31] Alright. I'll look for that.
[26:33] Yeah. But, again, if you're just going to do in and out, you know, file right, file left, you don't need that much stuff. You're if you're going to do conditional stuff
[26:43] and all kinds of interesting routing, then it might be worth it. So start with the free stuff, see where it leads you, and take it from there.
[26:52] Okay.
[26:53] Cool. Will do. Mhmm. Okay.
[26:56] Hey, Annie. Joe
[26:58] here. Again,
[27:00] it's funny. I just didn't upgrade this weekend from
[27:03] what was it? November
[27:05] 2021
[27:05] to this February 23.
[27:08] And I'm now running into that same issue
[27:11] with when I changed the account settings.
[27:15] Right? It says the current owner of the home folder does not match Mhmm. The file system owner, and it's a non root. But the funny thing is
[27:23] it's it's wrong. It's the the owner, the the UID and the GID and what the user password,
[27:31] it it is what's listed under the account.
[27:34] So I guess that might be a feature bug, whatever.
[27:40] It might be just overzealous.
[27:42] So I don't so when they release the feature first, they make it
[27:48] send the pop up on every single change regardless if you change the UAD, GAD.
[27:54] Then we all rebelled and told them no.
[27:57] So they changed it to be only when something is changed and actually need to ask. So I'm not sure if you are caught in the middle of this conversion
[28:08] or if you hit the bug.
[28:11] I'll open a ticket to support, see what they will say. I just really don't remember when they changed it, but I remember that when they put it first, it was literally every time you edit an account,
[28:22] it will throw you. You are changing UID, GID.
[28:25] Okay. I'm not, but okay.
[28:28] Yeah. I was just changing an email. I'm sorry. I was just changing an email on a phone on couple of different accounts existing,
[28:37] and,
[28:38] I hit it every time. So alright. I'll open a ticket. Yeah. You might as I said, you might be in this case where
[28:45] we were warning everyone just to be
[28:49] it was
[28:51] initially,
[28:52] we're doing it behind the scenes. People were not happy about that, so we decided to be all clear.
[28:58] And the way they were building that was every time they when they will call
[29:03] the CH own CH more CH own behind the scenes,
[29:08] they will throw the warnings, and they were going to wait too often. So there was something happening. I just don't remember which release that that was on.
[29:17] Yeah. And I tried to update it,
[29:19] but it's a non root. So and it didn't have the ability
[29:23] to
[29:24] to update it, I guess.
[29:26] No.
[29:27] When you're non root,
[29:29] we cannot do CH own
[29:31] because we don't run as a super user, the Java itself. But it also doesn't matter what UID, GAD you put. The UID becomes irrelevant.
[29:40] The GAD
[29:41] can be used for user classes, and from there for rules for access or whatever. Right?
[29:48] But
[29:49] because you are non root, it cannot touch the OS.
[29:53] Right.
[29:54] And you said you're on February.
[29:56] Right?
[29:57] Yep.
[29:59] Is the UADG ID of your users matching the UADG
[30:03] ID
[30:05] on the file system?
[30:07] Absolutely.
[30:08] Even on the even on the root,
[30:11] on the map. Okay. Right? It's it's right. So all the way back to the parent directory, and they're they're not nested. Each account has its own directory. Right? So yeah. Yeah.
[30:22] Yeah. Check with support what you just might be into the we'll always warn you no matter what because we're going to try that anyway even if we cannot stage.
[30:33] Yeah. I I mean, we don't change the email IDs and stuff like so it's not a major issue. It's just I wonder why it's asking me. So great. Thank you. Yeah. The reason they were asking to us because behind the scenes, they were doing it before.
[30:46] So they just pushed it up, but didn't clean it clean up the warning
[30:50] until a later version. And now in the May version, it's only warning you if it will actually do something.
[30:57] Okey dokey. Thanks. Yep.
[31:01] Okay.
[31:03] What else do we have?
[31:12] Hey, Annie. I have a question.
[31:14] Yep. Hey, Kamesh. Go ahead. Hey.
[31:17] We have a situation wherein there is files being pulled from SMB.
[31:22] But Mhmm. On the message tracker, at times, we see an failure
[31:26] saying it's outbound transaction.
[31:33] Okay.
[31:34] So what happens with the files that passing we're not sure as why it's got. It's transferring from inbound to outbound when it's in actually inbound file for us.
[31:45] Are you pulling from a subscription that is advanced routing?
[31:50] Yes.
[31:51] And if on this outbound,
[31:54] did you happen to look what the protocol is? Is it saying SMB or is it saying routing?
[32:00] It says SMB.
[32:03] Outbound SMB?
[32:05] Yes.
[32:07] It's surprising
[32:09] if my voice doesn't show you that.
[32:11] It shouldn't happen.
[32:13] So
[32:15] so what
[32:16] I would expect to see will be inbound SMB
[32:21] followed by the routing,
[32:23] lines,
[32:24] which can be inbound or outbound depending on what you are doing or both sometimes.
[32:29] If you're seeing outbound
[32:31] SMB,
[32:32] that means we're trying to push to an SMB
[32:35] or
[32:36] o.
[32:39] No. That that's the only case I can think of when you're pushing to it. Yeah. I'm just looking at my screen.
[32:46] It's the advanced road. What we have is that we are copying the subscription.
[32:51] It's trying to pull it's an auto schedule,
[32:55] and it's trying to pull inbound
[32:57] files from SMB,
[32:59] and it's assigned to a route that it is being
[33:03] sent a copy to an archive location that we have, and another one is sending to an external training partner
[33:10] or SSH.
[33:12] Is the is the r okay. Is the archive location SMB or is it just folder monitor?
[33:18] It's a just a folder monitor. Okay.
[33:21] We have sorry. S t s t. We are moving it to s t.
[33:25] Okay. S three. Okay. And you see the
[33:28] you see an outbound
[33:30] SMB
[33:31] and not outbound s three or outbound SSH?
[33:35] No. We are having say, for example, on a first iteration, it tries to pull from the SMB and it says one the first line would share inbound, and the next line would say outbound failed
[33:47] on the same SMB.
[33:48] And after a while, we try for the re retriggering
[33:52] it, It goes through fine.
[33:56] Okay.
[33:57] Do you have delete on success on the SMB? When you're pulling from the SMB,
[34:04] do you try to delete
[34:07] the file after you pull it?
[34:09] Yeah. Let me check it right away.
[34:12] Because
[34:13] On a on a normal nomenclature here, we just follow to pick the files from the source, and we delete those as soon as we have some people trying to do.
[34:22] Okay.
[34:23] When this is is successful,
[34:26] do you again see an outbound, or do you just see the inbound?
[34:30] Well, on the success, we see only the inbound transaction.
[34:34] We don't see any outbound information.
[34:41] Oh, okay. I can put theory.
[34:46] It doesn't make sense. So
[34:50] the only thing I can think of is that on error, we're trying to do some cleanup, and we're marking it at outbound
[34:58] for some reason, but we shouldn't.
[35:01] Oh,
[35:02] and it says SMB as a protocol. Right?
[35:05] Yes. It says SMB as protocol.
[35:08] Yeah.
[35:10] I wonder, do you have on the subscription,
[35:14] on error, do you have a checkbox to delete the partial file?
[35:19] No. We haven't enabled any such options.
[35:22] Okay.
[35:24] Because a deletion of a file will show up as outbound.
[35:29] Mhmm. So that was my theory that you we might be trying to do based on the acknowledgment that we received from the file system so that would turn into outbound. I understand that part. But we haven't done anything partial file deletion I mean, partial file upload deletions.
[35:44] I don't know, Kamesh. Honestly, that doesn't make any sense. My good feeling is that on error, we're doing something
[35:52] and something to clean up, and we're recording it weirdly.
[35:57] So,
[35:58] if you have a live example, get support to look into the system to try to get to the bottom of it.
[36:04] It might end the being just weird recording
[36:07] because
[36:08] the way the tracking table is written is basically just writes whenever there is a proper event for it. Right?
[36:15] So in order to have an outbound,
[36:18] there should be an outbound event of some type for another.
[36:21] We are doing outbounds
[36:23] in two cases.
[36:24] When a file is leaving the system, you know, the standard download push,
[36:29] but also on local deletes.
[36:35] So I'm wondering if it's not when that happened, is the size zero
[36:41] of of the file
[36:43] on the outbound.
[36:45] Let me just just
[36:47] a moment.
[36:48] Yeah.
[36:49] Because if it is a zero, it's much although if it's an error, the other one is also zero. So that doesn't tell us anything.
[36:59] But, yeah, you you will need someone to actually look onto the system itself and look through the server logs and see what is happening.
[37:07] Because it almost sounds like either we're doing cleanup and we're reporting it weirdly,
[37:13] or something else is happening
[37:16] that converts
[37:17] at
[37:18] it shouldn't be an outbound. You're too early for outbound.
[37:31] It's loading up.
[37:32] Also, I have one
[37:34] Mhmm. I have one another situation,
[37:37] Danny.
[37:40] I will yeah. Right now, I just opened that SMB error that we have.
[37:45] It says outbound, and it's the file file size is zero. K b.
[37:50] Yep.
[37:52] That might be a delete. If you click on it.
[37:54] Oh. Yeah. So click on it, go to the bottom of the page. What is the error?
[37:59] The error, it says connection to that server fail.
[38:03] And that's outbound? Yeah.
[38:06] On the outbound, it says, yeah, during transfer operation,
[38:09] The path and the file name, the file not found.
[38:14] And it says connection to server failed.
[38:18] Okay. It
[38:20] doesn't make sense. So
[38:22] get support to look into that.
[38:25] Either we're coding weirdly or we're doing something very strange here that I'm not thinking of,
[38:31] but I cannot think of what.
[38:34] Okay.
[38:37] Yeah. Can I see I can't Yeah?
[38:40] Go ahead.
[38:43] Are you using the shared file system on multiple servers?
[38:51] It's an SMB location onto on our
[38:54] network. Account. Account.
[38:56] You user accounts home folder
[38:59] is a shared
[39:01] mode or local file system.
[39:03] It's a file system. Not not a shared one. Okay.
[39:08] Yeah. Sometimes when
[39:10] we mistakenly
[39:11] use
[39:12] give us, I mean,
[39:15] a local system, we get a
[39:18] similar problem.
[39:22] We're showing an outbound
[39:24] on
[39:26] the tracking But it should have been inbound instead of outbound. That is failure.
[39:30] Okay. No.
[39:33] Actually, Kamesh,
[39:34] Rao is right. If you are using a if you are in a cluster
[39:39] with a local home folder, sometimes the server might get very confused because the file disappears midway.
[39:45] But still, it shouldn't be outbound.
[39:49] It shouldn't be outbound. Okay. No.
[39:52] Let let's get support to look into that, Kamesh. I'll be very curious to see under to see what happens here. It might be something in the SMB itself.
[40:01] It might be something in your configuration
[40:03] that, you know, some of your error handling.
[40:07] Okay. I don't know. Yeah. It just Okay. I
[40:11] mean, it
[40:12] it just doesn't belong to me, but
[40:15] okay.
[40:16] Okay.
[40:17] Thank you. I have another
[40:20] situation
[40:21] also. Sure.
[40:23] User is downloading the file,
[40:25] and it shows us outbound
[40:27] fail.
[40:28] Mhmm. On our outbound
[40:30] user
[40:31] and the
[40:32] it is more SSH.
[40:34] And Mhmm. Says file file transaction has failed. And it says transaction aborted.
[40:39] That transfer is aborted. But when I checked with the user or to the vendor, they have closed the connection successfully.
[40:46] And it is not showing any error on their end, it says as, outbound failed on our end.
[40:51] Yeah. What client is the customer using?
[40:55] Do you know the client application.
[40:57] Yeah. They're using a single application.
[41:00] Cloud application. Yeah.
[41:03] They are not closing something properly.
[41:06] So SSH is very particular SFTP is very particular on how things get closed.
[41:13] And if we don't receive the closing statement,
[41:16] we consider
[41:17] we we assume that something went wrong.
[41:22] Okay. So it's not showing up on either cases. They're not able to see that the error on there, and they see successful termination of No. Because yeah. Because they closed theirs because everything is okay, but they don't send the last package to tell us we got the file successfully.
[41:39] So we assume something went wrong.
[41:41] We're missing a package between the two of them. That's why my first question is,
[41:46] what software are they using?
[41:48] Their whatever they're using
[41:51] is either not configured properly to send things or is if it is homegrown, they haven't built it based on the RFCs.
[41:58] But, basically, in the s two world, unless we receive the confirmation
[42:02] on whatever protocol
[42:04] it's appropriate,
[42:05] we are assuming something went wrong because we don't know if they received the whole file.
[42:12] Okay. So Did you? Yeah. It's on their side. Okay. Okay. Yeah. I I was pretty sure it's on there, and we tried to justify that. But wherein we have been just getting this issue, and we I just try to reach out to customer support, access support as well, and they also did the same. And
[42:31] they were who gave them to provide us a a invoice what needs to be done to see their logs to show up that yeah. It turns up to when our good failure. Okay. I will I'll go ahead and raise that incident for the other issue,
[42:45] and I'll try to touch base with you in the next session if you have any. Yep.
[42:50] That that works for me. And I know we have a question in the chat, but Mark had his hand raised for a while. Exactly.
[42:58] Yep. I know. Mark, you're next, and then I'm going to catch up on the question in the chat by me Michael. So, Mark?
[43:13] Mark?
[43:14] Mark Lee?
[43:21] You're muted.
[43:32] Okay. I don't know what's going on with Mark. Mark, whenever you are back or if you have trouble talking,
[43:40] put something in the chat so we know what's going on. In the meantime, let me see the question in the chat.
[43:49] Questions arriving
[43:53] over okay. The questions from the chat.
[43:56] Files are arriving over SMB or folder monitor,
[43:59] but the outbound are delayed
[44:03] with a lot, like forty five minutes delayed.
[44:06] Michael, I know your headset is not working, so here is where I would start working.
[44:14] Make sure you don't have a very severe restriction
[44:19] on the SSH site itself on how many connections it can open.
[44:26] Because if you have already too many open connections,
[44:29] it might not be able to send through,
[44:33] and that's why they're delayed.
[44:35] And the other thing I would like to know is if it is only these files that are delayed or if there had been or if everything else was delayed at the same time.
[44:49] And I know you are chat only,
[44:52] so that might take a while to so everything was delayed.
[44:56] Okay.
[44:57] If everything
[44:59] so was there anything moving at all? Was or was just the server receiving files and nothing was moving?
[45:15] Because there are two options here.
[45:21] Okay.
[45:22] So everything else was fine.
[45:28] I so it's possible
[45:30] that so
[45:34] someone will need to look into the what exactly was happening. One possibility was is that because
[45:40] there were
[45:41] so many files going to the same place, the few first 100 or so managed to open connections.
[45:47] But then for whatever reason, they stayed open for a very long time. So all everything else went on a time out and waiting for them to finish.
[45:56] If
[45:58] if especially if it was happening only with this SSH side, the single SSH side. It's
[46:05] also possible that the TM was just blinking,
[46:10] as I call it. Basically, because you had so much volume going through it,
[46:15] it was
[46:16] a little stuck. So it was clearing to its queues.
[46:19] But
[46:20] because there were so many files to move through, there were so many things that had to go into internal retry that it looked like it's not doing anything while keep checking to try to find the file that can leave the server.
[46:34] I don't know if that makes sense to some extent.
[46:38] It's
[46:39] do you have enterprise cluster or standard cluster?
[46:47] Michael?
[46:50] Enterprise?
[46:52] Yeah. With enterprise,
[46:57] Honestly, without sync your your data, I don't want to talk too much about that. But my gut feeling is
[47:04] that because your it was such a busy time, your server is not fully tuned properly.
[47:11] So for a little while there, it was actually stripping to stay up properly.
[47:19] Yep. And you went to tuning as well. Yeah. So the way enterprise
[47:24] cluster is working is that if it feels like it cannot handle jobs, it will just leave them into the database until it can stabilize itself.
[47:34] So if you are not tuned properly and all of your threats are busy and so on, it does just doesn't have the capacity.
[47:40] Enterprise so standard cluster that's why I asked. Standard cluster will just crash one of the servers. The secondary will go down, or the primary will actually be usually first down.
[47:51] With enterprise cluster, the cluster will try to protect itself as much as it can,
[47:56] which sometimes mean leaving the events into the database.
[48:00] So and then we'll pick it them up when it's stable enough.
[48:04] So
[48:05] I would say some tuning
[48:07] is in order on your server. It's very heavy
[48:11] tuning, and we're not talking just memory and CPU at this point. We're talking about threat numbers. We're talking about
[48:21] all kinds of,
[48:22] you know, internal mechanics that we have
[48:26] so that you open it a little bit wider for cases like when that happened.
[48:34] It also
[48:36] oh, sorry. I was saying the there is a few parameters about thread sizes and thread numbers for the transaction manager for the event queue and so on. Those values will might need adjusting a little bit.
[48:51] Unfortunately,
[48:53] outside of the perform of the capacity guide, we don't really have a this is the best value go and use them. So it might be a little bit of a tie and,
[49:05] try fail, try again kind of situation, hopefully, without the fail in the middle.
[49:10] But we don't have a perfect numbers to to give you because
[49:15] everyone is different.
[49:19] So,
[49:21] if you had never done any tuning like that, run the numbers by support before you apply them on a live server and, obviously, test on development server, of course.
[49:33] But that that's the only thing I can think of. Either that or you hit the bug. So especially because SMB is involved.
[49:41] The plug in had a couple of problems in the past with not closing connections properly,
[49:47] which was causing this kind of blips until they timed out.
[49:52] I don't think you're on the latest version of the plug in. Right? Not just latest SD.
[49:58] And if you're on the latest, I haven't heard anyone complaining of it doing it again,
[50:04] But, I would keep an eye on it if if it happens again with SMB specifically.
[50:09] Get support to look into it. We might have missed something again there.
[50:13] But other from that, it's a little bit of
[50:18] it's hard to diagnose without looking at log files, obviously.
[50:21] But that's where where I would start.
[50:25] Get all the files go out finally even though they were delayed.
[50:34] Yep. Okay. Well, that's good news. That means the server recovered,
[50:38] which is what it was supposed to do. But I know that the delays are not nice.
[50:45] It's
[50:46] that's what it sounds to me like. Your server was a bit so overloaded.
[50:50] So one thing to check next time it happens. You know that we added the monitor for the event queue in the API?
[50:57] Look at how many events are in the event queue at the time when things are really slow.
[51:04] You probably will have a huge amount of events in there,
[51:08] which is not a problem on its own.
[51:12] Oh, you didn't. Okay. Go to the API. Let's see if we can do that. Hold on. I have a live server. I keep forgetting that.
[51:25] Okay.
[51:28] Hold on. Too many windows.
[51:34] API.
[51:44] It kicked me out. Not yet.
[51:47] So
[51:48] the
[51:49] bigger number of
[51:53] so in here on the events,
[51:57] you can do a get that will get you the events.
[52:04] And
[52:05] let's
[52:08] see. Limit
[52:11] limit 10,
[52:13] offset zero.
[52:22] See if I messed up something or not.
[52:25] So I don't have anything, obviously, but what it will give you is the total count.
[52:30] And this is actually the number we're looking for. So make a very small offset with a very of
[52:35] offset zero with a very small limit just so you don't drop thousand events. You don't want that. But look at the number of the events.
[52:43] What it will tell you is how many things are actually scheduled to be processed.
[52:50] And, again,
[52:52] that doesn't always denote an error. You can have in enterprise cluster, the biggest I've ever seen was about 50,000,
[52:58] and it was checking just fine.
[53:01] But if you have a lot of them, especially if a lot of them are pulls from a single site,
[53:07] they will check to them slowly.
[53:09] And the way it works is that the file the server will get to an event and say, oh, okay. You're a pool, but I already have a 100 or whatever your maximum is. So I cannot take you. Go for internal retry, which will send it at the end of the queue and so on and so forth. When the line is too big, we start getting back those that were from the back again, send them back over and over.
[53:31] Those internal retries don't count it as retries.
[53:35] So you can have thousands of them happening.
[53:39] And for don't forget that it's not just the polls themselves. The incoming ends might be also retrying if something is going weird,
[53:47] or they might not be processing.
[53:48] So
[53:49] that's the best thing you can do at this point. You you can look at the event queue.
[53:55] You can see what's going on. My guess is that on the very if your server was everything was waiting, it was basically
[54:03] that way.
[54:04] Another option, by the way, and it's specifically because you're coming from SMB and from external stores,
[54:12] look at the dates of those files. Is it possible that they were that something on the drives
[54:18] was
[54:19] postdating
[54:20] them. So they appeared files from the future.
[54:24] Because we'll pull them on an SMB,
[54:27] but once they arrive, we cannot process a future file. So if we keep the date of the file,
[54:32] who knows? So you have if you have some date discrepancies between systems, that might cause troubles.
[54:39] And that's pretty much all I can think of.
[54:42] Okay.
[54:43] You talked about most of it. So at this point, some tuning,
[54:48] and next time it happens,
[54:51] just look at the event queue, look at the server logs, see what happens, if you can get support live from the system to look. But
[54:59] I would not restart the servers. Anytime you restart one of those servers when they are dead stuck, it what it does is basically to flush the queue and start over to it,
[55:09] disrupting
[55:10] everything it had been doing. It's basically a balance. So what you did with just leaving it to run, and at the end, all the files went out is basically what it is designed to do. Stay up, handle as much as possible, as fast as possible.
[55:24] But that's
[55:26] yeah. I know. Everyone is tempted to restart. I can tell you. It will feel like it's helping because it will probably move about five files before it gets stuck again.
[55:35] And in the long run, something that we have taken an hour will take about four hours if you keep restarting it.
[55:42] So,
[55:43] what I do with this kind of stack servers, I'll just leave them for an hour to see what happens.
[55:48] Usually,
[55:49] and it sounds like it happens in your case, they will just solve themselves and start moving.
[55:54] But in all the in all cases, make sure you save the locks from that. Even though it resolved itself,
[56:00] get the locks to support to see if they can figure out the root cause of that. Again, it might be just load,
[56:07] so you might need tuning,
[56:08] but it might be a bug somewhere or something else happening.
[56:13] You know, that's the best I can say without locks and without spending the next three hours reading log files.
[56:20] Okay. Well,
[56:22] good. Okay. Mark, you back?
[56:27] Mark Lee?
[56:29] Nicole, can you try to unmute him?
[56:32] Maybe he cannot unmute.
[56:35] Sure.
[56:36] Let me try. I sent an email to Mark.
[56:40] Okay.
[56:46] Oh,
[56:48] okay. While we're waiting for Mark.
[56:56] Did you lose him?
[57:01] He is still here. Mark?
[57:05] Yeah. His his mic is disabled.
[57:08] Sorry, Ed. Go ahead.
[57:11] Hey. Hey, Eddie. I was just
[57:15] two other two other questions for me.
[57:19] The
[57:20] on the containerization
[57:22] front,
[57:24] we
[57:25] we were we were looking to obtain the the Docker images last year, and they said that they weren't actually,
[57:32] I'm sorry. Support said they weren't available yet.
[57:35] Are they are they still not available, or are they ready for public
[57:39] use yet? Oh. Do you know? We
[57:42] yeah. Well, that will be a two two pronged question.
[57:46] We do have some images at the moment. They're in,
[57:50] restricted availability
[57:52] because they're still looking into them, but we
[57:55] gave up on the idea of making SDS.
[57:59] It stands now container
[58:00] SaaS. Okay. It just doesn't work, not with our type of get databases. You know? Think about how we connect what we talked about earlier about databases.
[58:09] That's basically the antithesis of dockerization.
[58:11] Right? Right. Right. So instead,
[58:14] they were working on something that they that will be called SecureTransport
[58:18] Docker edition.
[58:20] Okay. Sorry. Cloud edition.
[58:22] No problem. Cloud edition,
[58:24] which will basically be our container setup, which will be a different type of application.
[58:30] They are projecting first built
[58:33] late next year.
[58:35] Late next year. Okay.
[58:37] So if
[58:38] you need something at the moment and you really, really need to go on containers,
[58:44] talk to your EA,
[58:45] and let's see if we can get you the current array builds. But if you can wait, I and you really need to go to cloud, I would say, to look for the cloud additional stuff, and they're already talking about stuff. They had a very good conversation,
[58:59] and they had advisory board actually last week on the summit for that.
[59:04] And I expect some meetings very soon,
[59:07] and I'll remind them that Dell is interested again
[59:10] because I know you are.
[59:12] Yeah. Yeah. But
[59:14] we we finally realized
[59:16] that what we were trying to do was to to fit a very square
[59:20] application in a very round hole that was three sizes too small.
[59:25] Mhmm. Mhmm. So we actually sit back and talk about, okay.
[59:31] If we cannot get the monolith in, can we do something a little bit more intelligent?
[59:36] Mhmm. And there aren't that many details yet on that either.
[59:41] There are some Okay. We know that the cloud edition ST
[59:45] behind the hook, it will be doing the same things, but, the way it will be positioned and put on the servers will be different.
[59:53] Mhmm.
[59:54] Okay. But
[59:56] for now, if you really, need the images, talk today or back to talk to support, see if we can get the GAs. But, you know,
[60:04] the the arrays, there is no GAs. There won't be GAs on the current releases.
[60:10] So
[60:11] Okay.
[60:12] Okay. Okay. Thanks.
[60:14] And then
[60:15] one other thing that
[60:18] so
[60:19] we have a team that that basically monitors our environment and and, you know, handles
[60:26] failures and and things along those lines. And and one of the topics that they asked us about was
[60:34] resubmitting failures. So
[60:36] if if, for example, right, we have we have a a bunch of files going to a
[60:42] server that's down for whatever reason Mhmm. And there's
[60:47] and that server then gets brought up, and we we, you know, we can resend the files. There's no way to, like, do, like, a bulk reprocessing.
[60:55] Right? Can't oh, you can't? How do you how do you do
[61:00] APIs.
[61:01] APIs. Okay. Yeah. But even on the API, it's restricted on how many can go in the same batch. It's basically trying to protect you from not shooting yourself in the foot. I think the, at the moment, the limit is 10.
[61:15] But, basically, when you go to where I am? Hold on. API. API. Let me share again.
[61:22] So there are two separate arrest submits. One is for a single file.
[61:27] You know, there's the one that everyone is usually using called the one from the UI. It's basically the same one.
[61:32] It's gonna transfer this. Hold on.
[61:37] So what
[61:40] so
[61:41] it's the
[61:43] oops.
[61:44] Sorry.
[61:46] Okay. I'm in the wrong place.
[61:49] Anyway, there are two resubmits. They might need to go to API one four. I don't remember the details on the top of my head. But, basically, there are two separate resubmits in ST.
[62:01] That's
[62:03] Okay.
[62:04] So but you said there's a
[62:07] there's a limit of 10
[62:09] you can do via the okay.
[62:12] Yeah. And then they can and then this API has both
[62:17] limit and offset, so they can cycle to them very quickly. You know, write it on a script or something and go zero to to nine,
[62:25] ten to 19, and so on, and they'll go.
[62:28] Okay.
[62:29] Alright. I'll look into that. Thank you. Yeah. And the reason why if you cannot find it in the two zero, I think it was one of the ones that didn't make it into two zero, but I haven't looked for a while.
[62:40] Look down on the one four API. One four. Okay. The old API.
[62:45] And that's basically if you cannot find that subject in the two zero, sometimes it might be worth looking into the one four
[62:52] Mhmm. Because some of the
[62:54] weird actions didn't make it up, and we had been tracking them. I I I don't really remember if it was in two zero. I know it was not so I know that when they released two zero, only the smallest submit as I call it, the single file I submit was on the API. So we had to put on one four.
[63:11] I literally haven't looked for e for a year and a half at least to see if it made it into two zero. I just put a note for myself. I'll check after the call because now I'm curious.
[63:22] Sure. Sure. But, yeah. But that's that that's what that's the best option. And the reason why we had to put the limit on how many files you can set at the time,
[63:32] let's just say that someone might have resubmitted 10,000,000 files.
[63:36] Mhmm. That might have happened.
[63:40] Okay. So we used to allow pretty much everything, but when we change the resume to allow resume meeting successes as well,
[63:48] that's when we screwed up.
[63:50] I see. You know? Someone actually rescuited their host server. They had 10,000,000 records in the tracking table. It was ugly. Yeah.
[64:00] Very ugly.
[64:01] So
[64:02] yeah. So that's why we have the limit. The other thing is, you know, the resubmit itself takes the same search parameters as the get. So they can do the get to search for them, and when they have their query properly set up, then they get use that for resubmits.
[64:17] You know, only errors only to that server, all that funny stuff. Or alternatively,
[64:22] they can just get the ideas and just use the API to cycle to them.
[64:28] Okay. Now and here is my regular warning because if you go to API, please, please, please do section management.
[64:37] You know? Log in once, use the same cookie
[64:40] to send all the rest of mids instead of opening a new connection with username, password every time.
[64:47] You know? Okay. You know that. Everyone that had been on this call, so know me on other things, knows that I'm always knocking for that. And then sent a log out
[64:57] just to clean up things a little bit. It's the admin UI. It's better than the HTTPDA, but still,
[65:03] let's not clock the server with session so that we'll never resolve.
[65:07] Right? Right. But, yeah, that that that's what is so let them play. Of course, give them for us a test server, please. Don't do that on the live server.
[65:17] Right. Right. Mhmm. Okay.
[65:19] I'll check it out. Thanks. Yes. Thank you. If you cannot find it, ping me.
[65:25] I'll track it down.
[65:26] I'm just
[65:28] I just don't want to spend the extreme minute time to remember where it was. I know it's there because I've done that. So Okay. Okay. Okay.
[65:36] Thanks. Okay. Come okay. Good. Kamesh, that's you. We we lost our other questions. So go, Kamesh.
[65:44] Yeah.
[65:45] Thanks, Annie. Two questions I have. Is
[65:48] there any plans for major upgrade
[65:51] from fighter five to fighter six in near future?
[65:55] We're in continuous development,
[65:58] and the current
[66:00] statement is that we are on five five continuously until we are not,
[66:05] which basically means no plans whatsoever,
[66:07] but we never know.
[66:10] Okay.
[66:11] And the second question is, what can be the maximum concurrent connection that we can do for an SSH file transfer
[66:19] for external vendors?
[66:22] Yeah. Yeah. That's a loaded question.
[66:26] My
[66:31] technically, unlimited.
[66:34] In reality,
[66:35] if you need more than about six to 800
[66:38] on a edge or server EU, I would strongly recommend you get another another server or edge into the picture. So
[66:45] the way
[66:46] I'm planning for size is every six to 800 connections require separate edge.
[66:54] Oh, okay. Yeah. Right now, we have 500. We are good for that. But on the subscription,
[67:12] max connection per host.
[67:15] So we cannot open more than a 100 connection to a single host name to start with.
[67:21] So a single transfer site, even if it is the only the same, if it is even if it is the only one on test host name,
[67:29] cannot Mhmm. Flow through that. You know, you have three separate levels for that. You have the on the site itself, on the subscription for pools, and you have the max,
[67:40] which is the global setting, the max per host.
[67:43] The smallest of those will restrict everything. So no matter how high
[67:48] you go on
[67:50] the site itself, it can never blow through the other two. So underhanded.
[67:55] Okay.
[67:56] We have a situation wherein some vendors do do have only the maximum con con connection that we could do is ten.
[68:03] So we have we're meeting ourselves to do the subscription
[68:07] for
[68:08] that at ten.
[68:09] But still, say, for example, if we have 11 files or the 10 files, at times, the nine files goes to one file phase. So Yeah. Yeah. So if so if you have 10 connections,
[68:21] set the value to nine.
[68:23] Let ST make a mistake.
[68:25] Because we have so many confident connections,
[68:28] I would always account to 10 to 20%
[68:32] shine chance of us blowing through the numbers.
[68:35] So if they allow 10,
[68:38] I put either eight or nine in the field just to be on the safe side.
[68:44] Good. Thank you. So I I did that same right now to have that to go through. I made it eight for 10 against what all they have given. Yeah.
[68:52] So just just a little lower. You know, because we're doing calculations and we have a lot of things going in parallel and especially when you have a lot of small files or when we're pulling,
[69:02] there is a huge amount of events that literally ping the same time. So if you have three pull files in three separate threads at the same time, checking is there a connection for me, chances are two of them will step on each other toes. We're trying to resolve it in a lot of ways,
[69:17] but plus one, plus two, you know, 20%
[69:22] on the smaller values especially,
[69:24] it's inevitable.
[69:26] So I I would do minus two if you can. I mean, they won't move that fast anyway. So Okay.
[69:33] Yeah. Thank you. Thanks. Yeah. So that's that's what
[69:38] basically,
[69:39] you had two files in parallel asking, am I good to go? And they were literally in parallel. That that's the problem with multi, multithreaded
[69:47] systems. Right? We're a lot better than we used to be. That that I can tell you.
[69:52] Especially now we're up in the coherence layer and not locally on the servers, but still it can get a little slower.
[69:59] So,
[70:00] that that's what you need to do. Just go a little lower.
[70:05] Another thing, Kamesh, by the way, you might want to put the
[70:10] setting on both the site and the subscription. So both of them can enforce each other.
[70:16] So even if one of the checks fails, the other one might catch up.
[70:21] You mean to say that on the transfer site?
[70:24] Yeah.
[70:25] Okay. Oh, sure. Yeah. Top of the transfer site, you can put the same number over there that you have on the subscription.
[70:31] Technically, it sounds like double dipping, you know, putting the same setting in both places.
[70:36] But in practice, because of timing consideration, it's possible that one of them get checked faster or you know, you're beating a parallel action.
[70:44] That
[70:46] that's what's going on. And and sometimes we just no matter how many times we lock and so on, we either need to slow down everything,
[70:54] or we can just account for a little bit of a differential.
[70:59] And yeah. It that's why keep a little lower. That's all.
[71:05] Okay.
[71:06] Good. We're almost Yeah. Thanks. Yep.
[71:10] Perfect. Okay. We have ten more minutes. So we have time for about two more questions, maybe.
[71:17] Anything else? Anyone else?
[71:24] Annie, Mark replied. It was a an error. He apologized.
[71:29] That's okay.
[71:32] Okay.
[71:35] Anyone else? Last call for questions until six weeks from now or thereabouts.
[71:42] If not, couple of housekeeping
[71:45] items while I have everyone. So as you know, we just hit our summit last week. It went very well,
[71:51] as some of the people on the call that were also in the room will tell you.
[71:55] I actually did the presentation of very interesting ways of how to use ST Smarter.
[72:01] We were thinking of doing them piecemeal
[72:05] on these meetings. Instead,
[72:07] there will be another user group in a few weeks called Smarter Secure Transport,
[72:13] and I'll just, have the presentation for the from the summit. It's basically advanced routing trickery and fun stuff with, advanced routing.
[72:21] So if you see that one,
[72:24] that's what this is all about.
[72:27] Okay. Last call for questions.
[72:36] Hi, Annie. This is Paula Zuback with Chubb. Can you just tell us again where the free training
[72:42] is? Oh,
[72:43] go to Axway University.
[72:46] Okay.
[72:48] And you should see listed whatever is available. And I actually have the trainer on the phone
[72:54] at the moment. So, Ian?
[72:57] Yeah. I can't remember the exact link. I I can send it to you and put a chat. But there should be a, a learning path, and that will give you all the free courses in there. Okay. Great. Thank you. Thank you, Ian. So Ian, Percival, is our
[73:10] is a person responsible for all the skilled transport training. So
[73:15] that's why he,
[73:17] joined our meeting in case we have questions about university. And Zhihu, we had. So, yes, they built the
[73:27] learning parts that should help you?
[73:30] Okay. We have question in the chat. How to add their high volt files to separate parts on different file system?
[73:38] Yeah.
[73:40] I can I'm sorry. I'm not trying to I'm not going to try to pronounce your name.
[73:45] Vish Vishnudas?
[73:46] Something like that. I'm sorry. I'm bad with names.
[73:51] You cannot do it with the archiving itself, so you cannot do it on the subscription level.
[73:57] What you will need to do is to create a folder monitor site that points to the other location and basically do an outbound transfer. Push the file into the system at the end of your route.
[74:10] So instead of making it an archive on the subscription page,
[74:15] you just add a last step inside
[74:18] of the routing itself,
[74:20] which is to push to the folder monitor.
[74:23] And then on on the push itself, on the center partner,
[74:27] you can change
[74:30] the,
[74:31] upload path to put it wherever you want it to go. Another option, if you don't want to do a push to a folder monitor,
[74:39] is to create a special SD account for archiving and put the and put the archiving location as the home folder of this account. And then you do the same as the center part, which is that you publish to account,
[74:53] and you publish to whatever folder of that account.
[74:56] Publish to account might be better in some cases because it can create folders for you when you drop the file and so on, But it will also create the dot STFS folders on the receiving side.
[75:08] If you don't
[75:09] mind them, then you're good to go.
[75:15] Thank you.
[75:16] Okay.
[75:17] Yeah. It it's it's not straightforward and
[75:20] but but it's cleaner. It will also show up in your tracking table, so you have better control. I personally like using the publish to account for two reasons. One of them is it doesn't count as an extra as a transfer, outbound transfer for the license for the count of subscriptions,
[75:38] but also because it can create the folders for you. And you also can choose what would happen if this file already exists. So if you are trying to archive the same file, you can actually actually put
[75:47] a timestamp or, you know, you can make sure both of them survive and so on.
[75:52] But you need to deal with the dot dot STFS folders.
[75:56] So you know? Because it's a whole folder. Right? But that's pretty much it. And Ian has the link for the,
[76:06] base for the training in the chat window
[76:09] for anyone that needs it. But if you don't catch it from there,
[76:13] university.xua.com,
[76:15] look for the
[76:17] learning parts
[76:19] or learning plans,
[76:21] and we have things there.
[76:24] Okay.
[76:27] Any other questions?
[76:29] Anyone? We're almost on time.
[76:33] Oh, we have five minutes left.
[76:35] If no one has anything,
[76:37] thanks everyone for joining us this morning, afternoon,
[76:41] whatever time it is, wherever you are.
[76:45] And And
[76:46] Yeah. I have one I have one final question. So you said that you're going to send us the link for the recording.
[76:53] So how do we access that,
[76:56] recording?
[76:58] That's for for this meeting. That's what Nicole will tell you in a second.
[77:03] That's her. Okay. I talked and I answered questions.
[77:06] She's doing all the stuff around it, organizing, sending links,
[77:12] and telling you what's next.
[77:14] And, again,
[77:15] as I said, there will be also another one. We'll