GitLab Geo Group, 10 Dec 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Geo helm chart demo (part 1)

Description

Recently the helm charts to deploy GitLab Geo in kubernetes were merged. In this demo Toon Claes (Geo Senior Backend Engineer) is following the documentation set this up.We ran in a couple of errors, so this part 1 only goes up to the end where the license is added to the primary node.

Documentation: http://docs.gitlab.com/charts/advanced/geo

Issue: https://gitlab.com/gitlab-org/gitlab/issues/36965

A

B

Okay, here we go so let me see if the setup of the secondary cluster is still going. Okay, it did again complete, but it again gave me this error, which is strange, but as far as I understand, we can ignore that.

B

So at the moment we have two clusters: I already set up two VM instances for setting up the post class databases. There are those in the same zones as the clusters for the kubernetes and I. Think I should be good to go for yeah.

C

B

Starters now so I've have. The issue here were, with all the information filled in about what what, where the machines are so I can check these boxes. Now.

B

It record ok, so the documentation. That's this one I.

A

Am seeing the mystery critical step is choosing in beverage at temperature of your choices, yeah.

B

It's behind me, don't forget it.

B

Thanks Jason, ok, you're more than welcome.

B

The documentation, so what did we do so far, so we're using external PostgreSQL?

B

So it's not in the shorts support application stuff like that I guess it just omnibuses, so it should be, should be supporting test setup omnibus instances. That's what I've done already I'm, not sure. If, for this I need some extra firewall rules, maybe we'll need to check that out later on other than that, so I've use a operating system and I have installed two omnibus package on there. Without an external URL I've set up two clusters, which I've done just yet both English using HTTP I'm, not sure what this step additionally is.

A

Effectively, they both need to be able to reach each other over port 443. Can you deploy the charts in gke? This should not be a problem. Okay,.

B

Great, that's what we have so, let's move up.

B

Access to persistent storage mean you're not required if using external. Yet these I wasn't sure what I'd be sure it should be doing with I did not set some set up any Gately or mini or Redis I'm, not sure. Maybe we can offload it also to the Omnibus installations or the charges doing it, I'm, not sure where they are supposed to resign. Sure.

A

The this probably could use slightly rephrased and but effectively it's saying use the things that, if you're going to use maternal, you need to make sure that your clusters can access if you're deploying it as a part of the chart as we're doing in this demo, then again, this is one of those things that's taken care of by the chart and you don't have to be concerned.

D

D

Turn if you're talking, you're, muted.

B

Strange I'm not sure how that happened: okay, so yeah, so I did collect information and I all wrapped it up in the issue, which was should be linked in the agenda item. Also, so that's the first part, then we can set off with setting up the primary database on the Omnibus instance.

B

So this is the SSH access to the primary.

B

And we should be installing this into the keep the power P.

B

We need to replace several items: external 0, L.

B

E

B

Bit confused because I didn't expect this host required to be publicly accessible like from the outside.

A

External URL is actually throughout many other pieces. It's not strictly about the automated configuration of HTTP access. This won't be accessible through 483, but rather external URL get populated into things like the geo node name when not specifically defined oh yeah.

B

That's true, so what's your more concrete suggestion here.

A

Set it to whatever you expect the external URL to be of your primary or any of these cluster.

B

Set it to whatever you want to be primary, could we need its cluster okay? So that's the same name throughout like the whole installation, yeah.

A

If you're, if you're geo primary, is going to be get lab change that place, then you would enter that here.

B

So, let's use the same, so TC cube primary that Colgate laptop ml yeah. We need to add that to the DNS later, but fine.

A

You may want that to be HTTP yeah.

B

Okay, wasn't sure.

B

Make sense what else did we need to customize the user password hash?

B

So that's the this one.

B

Again and the PostgreSQL and the five sitter addresses can be updated to be a list explicit IP addresses yeah. We don't really have.

A

It set to zero zero, which basically means anybody can access if they.

B

Yeah so it will will work, but it's not really like safe to do it like this right for.

A

The purposes of a demo it'll function in production, don't do that yeah.

F

B

Okay, that's it so place the continent, kid lab or be reconfigure set replication password. Let's do it first, safe.

B

B

That was fairly quick, so let's set application password yeah, also for the purpose of the demo. I've just used very simple, password and I have to use the same one for every host for every database. It's also not really recommended, but it's pretty simple for this demo. Okay,.

B

B

That's retrieve the primary database public certificate.

B

B

Okay, so the primary database is good to go well now we can deploy the primary key Oh kubernetes cluster, we'll start from this example. Configuration.

A

There is also a lengthy actual example: yellow file. If you choose.

F

It's in the charts.

G

A

Give a a curl to get the wrong key. Sorry.

B

A

Can clone it or you can do a curl to get the raw output or you just copy and paste the parts for the documentation? Let's do it.

B

Yeah I have it here so domain.

B

Okay is it's like the whole domain, or just in.

A

This case, it would be the domain part. You want to specify this host names. You'll have to specify those and separate. This is just the actual domain. I missed what your host name was supposed to be able to primary.

B

Where did I put that again lost this one.

B

So is it this one or not.

A

Yeah, so what you do is just have the lab, NL or ml. There.

F

A

E

Yes, you're actually going to want.

E

E

Whatever your actual external URL.

A

And then I friends need to change those hosts for any reason.

H

A

You have full control of the domain. This is no problem.

B

Yeah but sure I rather use like something like this and then then it's.

C

B

Redis would would be created, or is it.

A

Yes, so you're correct there, so you wouldn't get ready start primary, but you would get get layup that primary deal. I got em out.

B

Okay, I'm just.

A

B

Yeah I just want to make sure that my primary are my secondary, aren't using the same host names or the same domain. Sorry, which is good! It's good like this. You should work. Yes, sorry Jason! Yours, sometimes your sound, isn't great, not sure if it's my side, but your your sound like.

A

That it's me and airport air.

B

A

Airport, sound sorry, okay,.

B

For the PostgreSQL host can I should I be using an IP address here or should I be setting up this as or should I be setting up DNS records.

A

If you use a host name, that in DNS record will be required, you can use an IP address directly. Either one will work as long as the TCP stack.

B

Sorry I didn't get that.

A

If you don't want to wait for setting up the DNS, go ahead and use the IP yeah.

B

I was planning to use IP, it seems like simpler end and it can be an internal address for gka I would assume so.

B

B

No I'm not sure with secret CEO.

A

Eight cubes all secret before boy.

B

Sorry, Jason I cannot turn and understand you.

H

If creation of.

A

That secret is documented.

B

That's the well yes,.

B

They, maybe it just have look back to the talks.

B

We need to create a secret containing the database password for a short consume, replace password below with the pressure.

E

B

B

This is not I expected much more.

B

Aha, okay, that's from the door! Yes, so I need to read before this and are probably know this. One I run this secret co-created I, wonder where it was created exactly.

B

What else do we need?

B

We need to create a secret, updated configuration, correct values for a global hosts domain pieces. Baskerville host right also come configure any additional settings: SSL TLS external Redis, external object, storage. We don't use it. So that's fine, deploy the chart using this configuration.

B

That's in the examples geo primary.

B

Let's try this yay.

B

B

Hang tight while we grab the latest from the shorts.

A

Yeah that we need to add they get library. So that's why it's I.

B

Did I miss in preparation, then.

G

It might be very.

B

Sad I remember.

G

Seeing earlier I thought the did that.

B

Okay, step I know.

A

The bootstrap scripts to poise the cluster and everything, but it has nothing to do with helm at this time.

G

A

A

G

Going to assume you asked me to help in either repository, because your audio is.

G

So yeah, if you do a.

G

B

G

Get your beverage.

B

G

I do I have my hot beverage of choice, but I just haven't drank enough of idiot how many people and get.

G

Actually, if you, this is right top of the fold of our sharks, docks, which is.

G

C

G

A link in zoom chat to our charts, docks.

G

Okay, you can see the commands under get lab version. Mappings are exactly the first command is exactly what you need to do to add the repo to our.

B

Was there anything else, search the surges just to verify whether it worked I, guess yeah?

B

So now the upgrade should work.

A

He's missing the having an email address for the certain manager, because we have set up for HTTPS yeah.

G

So this is due to the built-in- let's encrypt integration. You need to give it an email address for that, so that should be added to the dots, but you can do it on the command line as well.

G

So if you add to your upgrade to command, if you do a dash test that.

G

Sir manager, if you were email, oh.

B

G

And then give it an email address like.

B

B

This looks cool.

B

State is deployed.

B

Resources all right.

H

Have you set up the ENS.

B

Have you set up the DNS I'm, not sure what what do you mean by that yeah I probably need to set up something in DNS I'm, aware of that and I saw somewhere I needed to give me this DNS access, but I missed where I should have been doing that.

B

B

Just joined guides.

B

Is it generating that for me or should I should I have told it somehow to create these DNS records? For me,.

G

Automatically for you, if you use the static IP as soon as you're, going to manage that.

B

Sorry, I'm not sure what I should be doing here now.

G

So I think we need to create some DNS records for this. In this case,.

B

So it should be creating for this IP address a DNS records or I.

G

Know if you do a command line, if you do a cute control, get ing or get SVC. Sorry.

B

G

H

B

No, it's don't like P address. You.

G

Should have services from there should be other services there? That's not what I expect to see it all.

G

Can you do it get pods.

G

Are we on the right communities, configuration I, mean I, guess so, you're doing to kubernetes clusters right now, right, I've.

B

Created two: yes, so.

G

We might be on your second cluster at the moment.

B

Yeah, that's what I'm missing like how disorder this you control command array, so we can go through the Google.

G

Console, if that's easier, this point in time, rather than walking through how to manage the kubernetes configurations, because that's a whole nother story.

G

So service isn't Ingres, sorry, so I'm, better the coop control, so you'll have to bear with me a little bit but you'll see in the left hand, side yeah. Sorry, that's one.

G

And then you can in the filter thing you can do a cluster : and the name of your primary cluster that'll reduce it.

B

B

Like this, okay.

B

Alright I've done something wrong: I guess you.

G

Might have deployed your secondary cluster because when you, when you use the bootstrap script, it sets up your cube control, configure automatically to use that cluster. So we might have and now deploying your primary to your secondary, close I.

B

Was afraid something like that because I saw them secondary here, but not yeah.

C

B

In our surprise that also this doesn't return anything so that may.

G

Be the best alone, your primary configuration for free cube intro. So it's alright, it's just a name. It's you know calling it secondary doesn't necessarily mean it has to be the secondary. So.

B

G

Can work with us.

B

What this is looking like, if you would like deployed in regions, wouldn't that be the wrong region or if yeah.

G

We want more careful with this, so this might be something we wanna, maybe call out on the docks that deploying the clusters might mess with your kubernetes configuration so to make sure so you might want to put in some notes to make sure you're on the right cluster before you run commands, but for purposes of today we don't need to remove this and yeah.

B

Although I'm not sure so we wanted to set an external IP address or a DNS records right.

G

So if we go back to the cloud console, we need to find the external IP address. It should have dynamically assigned your external IP address, so we need to look at the services for the cluster and we're looking for a type of load. Balancer.

B

Mm-Hmm so let's say, yeah select secondary go ahead, so these are all in secondary registry. You notice.

G

There's the load balancer right there.

G

A

G

Address for the gitlab gon jet engine X fingers controller.

G

Ccc, the third column is load.

B

G

Yeah so and there's the IP address that we need to create DNS records for.

B

G

Then we'll want to do a wild card.

B

So that's primary then can I do a wild card like this I'm, not sure I. Think.

G

G

B

It select yeah I did okay, it's not very clean for a reservoir environment, but for a demo.

A

B

Yeah, okay, so we can continue with the documentation. I guess we.

G

Will probably want to rerun the helm command with domain set to the primary dot I.

G

Don't think we have domain set in the example settings.

B

G

Yeah, it does narrow mic, so we should be alright.

B

B

Get lab geo: this was it okay, so I deployed this one. I did run this command, wait for deployment to complete login to get lab and upload your key lab license file, so I should now be able to yeah. If my DNS records come true, yeah.

G

Since they're new, they should be fine, we're.

B

Able to access this URL yep.

B

Of course, I can name cannot be resolved.

B

Yeah, probably not a good idea, but well something is listening. That's okay,.

G

That is resolving for me from home.

B

So you're you are able to access kit lap on that yeah.

G

If I did a day going, get lab primary go, get lab dot ml I get the IP address. We expect.

B

That's from my aunt.

H

Okay, huh well, the.

B

Water wildcard is working yeah.

B

Well, the TTL is set to five minutes, so won't shouldn't be taking longer than that, but.

B

Yeah Firefox probably also won't be able to get it now.

G

It's always dienes is.

C

E

Yeah, that's fine.

A

You can always.

B

I'm still wondering what I did wrong here or what I should be doing for the secondary.

A

This comes down to ensuring that you change the control concepts.

B

Yes, so that the bootstrap script did did set that context. For me, yeah.

G

So when the standard is for the secondary water make sure we update the context to switch okay, which is just a poop control, config set context and then TC g oq primaries will want to install the secondary.

G

I'll put that in check because come.

B

On this is ridiculous,.

B

Refreshing ten times doesn't make it go faster, or does it.

A

B

Not going to get happenin upload your license, let me quickly browse to what next command should be.

C

A

We have made this simpler, its merge earlier today, so 12.6, possibly the next or 12, by.

A

Resolving that.

B

Yeah it's resolving, but it's not healthy, I'm, not sure if this is a step forward. I think.

G

B

H

C

You haven't check the migrations, that's good cause.

C

G

Yeah we go back to the Google Cloud console.

G

And if we go to work loads.

B

Work notes: okay,.

G

Migrations pods are pending I.

B

See that I don't know what I thought that.

G

Probably means we're having problems talking to the database.

G

If you click on that.

B

I'm wondering why this Oracle geo, because just doesn't doesn't it's not related to its thinking, it's all secondary right or is it right.

G

So we leave the helm upgrade to come and we gave the release a name of kit lab geo, and so that's every part and deployment it's gonna create. Is that means it'll be deployed with github geo as a prefix to the name for everything? Oh yeah.

B

B

Okay, yep exactly.

G

So now you can.

B

Name that tune.

G

Super geo cluster and then every bottle they tune super geo, cluster, gotcha, postcards and cetera, et cetera. So that's how, if you click on the pod name, that'll give us more detail.

B

G

Workloads Google Cloud means pods for the most part.

H

Paul potts are pending.

B

Fault initializing.

A

Probably a problem so we're going to want to check the container logs.

G

Clock is very ominous.

A

To control logs might be faster, yeah.

G

If we want to switch back to your switch, your contacts like this you'll need to switch your cou control con context. First I put some some details in shed, but you run a config get contexts.

G

Fuu, what is the Fuu? It was just an example just Arkansas set whatever whatever context you want to set it to, but to get the name of the context, if you run the get contexts command, it's.

B

Nice, someone I'm not sure what but it capitalized key yeah. That's not me. I swear yeah I assume so so we need like this or is it yeah.

G

So yeah, where should that make sure that's that I mean it's as it is, but we weren't getting correct output, so yeah.

C

G

G

Iris, that context is singular.

G

All right now and then the lot the pod name. So if you do a or if you do uh L app equals migrations, sorry app app, a PV equals migrations.

G

All right, let's go back to the pod name this. So if you just control logs, then get lab geo migrations, then you should be able to tap complete. Oh.

B

G

If you set up your oh.

B

G

Okay, so queue control get pods.

B

That's one word two.

G

Words, sorry yeah.

B

G

All right so cute control n get lab, get pods.

A

G

We go and now so the.

G

And we'll need to give it and get lab again.

G

There are ways to make that default. All right. um Let's see.

G

So we have in it containers on this, so we can see. If you do a see, we can patch the different containers names to look at the logs for them.

G

We go back to the console window. Okay,.

E

G

To get the container names because I forget which containers we have with this.

G

Container so configure I'm guessing is the one we want so go back to the console now for the terminal. Sorry, yes, and if you do so, the dusty configure and in the pod name.

B

huh So again, this one yeah.

G

Yeah, what Jason, just posted in chat, will be useful.

B

G

The if you get instead of migrations hard to figure with the pod name.

B

And so again, and this yeah.

G

B

I'm happy it makes sense to you because that's for me so the last line there failed mount.

G

Secrets do not found.

B

Uh-Huh probably ended up wrong namespace or.

G

B

Yes, so cute control.

G

B

B

No, probably like oh yeah, that's what's wrong, I guess so.

G

You can think you could change the name space. So if you do a Q control, edit secret key Oh.

B

Edit secret he'll yeah.

B

G

The name space to get loud.

G

Alright, maybe you can't do that.

B

Doesn't yeah hey.

G

Kent yeah, so we just we just need to create it, recreate it in the correct name, space.

B

How did I do that again? I.

G

Think you did it with a cute control, create secret.

B

G

We add there mmm exactly.

C

G

Right so now the pod should sell for cover all.

B

Right should sell for cover yeah.

G

So I should try again and see that secret exists. So if you do the describe pod again command again, we're probably hitting a back off threshold right now, so it might take a little bit to make it move forward. You can always delete the pod and then the deployment will recreate it or the job should recreate it. I think mm-hmm.

B

Makes sense but I'm not sure our.

B

Dogs, well, somebody is happy.

G

I'm, actually one minute late for a call with Steven, who is also in the call already so.

B

Well, yeah, whatever, if you.

D

You know that's.

F

G

F

A

And we're already rescheduling mine too, so it'll.

D

Be out no worries guys. This is this is super interesting.

B

So shall I remove the doors just to sell. You wait.

G

Just check again now we could have recovered while we were waiting. I, don't want to delete it. If it's already running yeah I think we can delete this then so, instead of described pod, if you delete pod.

H

This is interesting, hey.

A

So how this works is when the pod is removed, the converted, YZ job will recognize. Pod was removed and recreated because it hasn't been completed yet.

B

G

B

So it hasn't another random year and it's run.

G

Out running and now, if we do logs on that, our cube control and get lab logs, if you do F- and you don't want to configure this time- oh.

G

Sorry, no problem and if you add f as well, it'll act like tail.

B

So yeah this is, this: is door disorder locks from the configuration phase right, yep.

F

B

F

B

More operational, interesting.

B

Phases- maybe not the right word, but.

A

C

Actually great.

A

Demo, because it's showing us the the level of familiarity that we have other people don't and it's exposing the knowledge. That's not necessarily available for everybody on how to do some of these tests.

B

Yeah yeah- that was my purpose for this demo, also like to an unbiased demo. So I didn't want to do too much preparation for this and just try what see what happens.

B

That's right, if this is supposed to take so long.

G

It may be a deer database or where we can tell the logs there and see it's getting any errors.

C

G

B

There's not much happening not.

G

Much means there's no errors, at least.

B

Okay, you would have to expensive suspected someone was rejecting connections here or something like that. Yeah.

F

A

H

Super speedy to start.

B

Yeah I know it's a yeah, it's a pain.

A

E

Working to try.

A

E

Boot, camp patchy so.

A

C

A

B

Well, that will be interesting. I know we had an effort for bootstrap in the past, but I'm not sure what trying to are on there.

A

It actually works perfectly fine on omnibus the problem is in the containerized ecosystem and the way boots nap actually checks. What version of ABI is running against and we're working to catch that to be correct so that we can actually rebuild the cache and distribute it so that it's not locked to a very specific kernel version and compile time.

B

That's an interesting exercise.

F

H

G

So do you have a firewall on your database servers the only thing I'm thinking of us? If you do, we might not have, did we open up access? They switched clusters.

C

A

Would be my concern? Yes, we configured the internal IP address and it can't see the internal IP address across the zones.

H

Yes very possible: it.

B

Was my purpose to stay in the same zone, but that went wrong at some point.

B

But something should have happened here now: independent of connection errors, I assume.

B

Is there a way to to test the connection from like a database yeah? That's not that's, not correct I guess we can do it to.

G

The task runner.

B

G

If you do, if you go back to the your terminal.

G

So you'll see the pod name is to get a geo task runner this one yep.

F

G

That's not actually good, not gonna, be running, is it or should be, can do it get pods just to see this current state.

G

Last night.

A

If you kill that pod, it should restart, it was probably well if it's in the middle of an it I would do it described on that pod see if it's coming up.

G

Okay, so yeah, if you kill, if you delete the pod, then it should. It should launch.

A

Yeah and I think this is what you saw could.

G

Be yeah that is just waiting is never retrying. No.

A

Get lab geo geo PostgreSQL passes. I think this is the bug you saw when, when going through, mr 1055 I.

G

Was look at the finish line worth and secrets? Do not found sacred state level, yah, so yeah. We.

A

Had saying the key is not found inside of the secret: oh yeah,.

G

Right yeah: this is a yeah. This thing yeah.

A

This is, this is a bug in the 2.5 with the inclusion of geo. The task runner is trying to have both passwords on the primary on the primary. It never needs the secondary database password, but we didn't pertinently introduced. This is solved with 1055 merged.

G

In the meantime, we can just edit the secret and add that key and we should be fine or.

A

I think this one actually wants a separate secret, but doesn't matter what content you put in it yeah.

G

It works fine for me to just add the added to the existing secret cuz. It's looking for the geo secret with that key. So if you do the cube control, edit or control and get lab ready to take your deal well,.

A

Actually, it may not depend what his helm configuration is that we didn't set you piece equal password secret. They would look for the template, a name which is that one present here right.

G

If you do the control added secret, kill.

G

Sorry, if you do the cube control and get lab edit secret geo.

B

G

If you create, under the data key, we need to create a.

C

G

Name whatever name the give a geo geo post ghost password, it needs the full get lab geo as well. That's where talking for.

B

Yeah, it can be same yep.

G

Give me the same.

A

It can even be completely bogus so.

A

We may have to create a secret by that name, be enter.

G

It should it's looking for it in the geo secret, so.

B

Sigh delete it to apply that or sure.

B

What can go wrong.

G

If we're doing it right, it should always be safe to delete pods. You might get some downtime while the service is restarting pods, but it shouldn't like destroy anything.

B

F

B

Looks better No kidnapped you're.

G

Looking for the full secret, all right.

G

So if we do a just create an empty secret with that name.

B

With this full name, yeah.

B

Like that, yeah.

G

B

Then again later.

G

We might be soon enough that it's gonna retry quickly, so uh-huh there we go.

B

Okay, what are we doing again so.

G

Now we can actually do it actually.

I

G

If you want to do a bring up that command that we just ran, we're gonna edit it a little bit. If you do change the describe pod, exact I, T.

B

G

And then, at the end of the line, do a dash dash bash. That's your space badge.

G

And now this is a running shell on a pod in the clusters that we didn't use to try and debug our database connectivity. It should even have psql, so we can use that.

B

Contrasting I, don't.

I

Remember if we put a wrapper in for lab since you for we did not.

G

But you can do it, we do have a gate lab rails in here, I think so. Yes, we do get lab rails, console and see. If that does anything.

B

And then we wait another hour.

G

If we get bored, we can always just do some manual commands with psql.

B

B

Stuck so I open another one, but that I knew ducks sorry just.

G

Control-Z work and then kill percent one.

F

B

H

I've never done it like that.

G

Yeah and your again, oh.

A

G

B

C

B

Yeah I need to host and stuff like that yep. So.

B

B

G

Capital? U and username.

B

That's kid lab 18 course: yep.

G

And then it should prompt you for the password at.

B

Us it should any if.

G

It could connect.

G

Yeah, it's not connecting that means that looks like yeah there's yeah.

B

That makes sense, there's.

G

No route so either firewall or just no internal route between network zones.

B

Just like this well doesn't matter, yeah.

G

It should be feeling psql should fail a lot sooner. Uh-Huh.

B

G

Actually connect to the database, so this tells me it's not able to connect.

B

Well, that work no I can't this.

E

A

The cheater method is try and curl.

E

It cool or open SSL client connect.

B

Open there's, no s client like that. So.

G

G

That's underscore client I think right.

G

So I am saying is you're using the internal the private IP address, so there might be on different networks. We might have to try switching to connect to the external IP address, so that was in the console window. There. Yes.

B

So this is IP I filled in here right.

G

That's the private IP address that might not be accessible from the cluster, but the external IP address should be accessible.

F

Right that that's.

B

G

So go back to the task, runner pod. We can verify it before. We update the configuration to use that.

B

H

G

They need open firewall here, yeah.

B

G

You should be able to do that through the console. I.

B

B

No, it's in there I think.

I

You just go to the instance itself.

B

You know should add in firewall, rule yeah standard or add a tag and then some network attack. What is it.

B

Where is it firewall.

B

Geosync right.

B

B

I've done this before, but I can't remember exactly.

B

Not yeah they're, like.

G

Hey we can just create a temporary one and tell you it when we're done.

B

But how do you connect her rule? Tour I would.

G

Just I'm slightly paranoid, so I would recommend not opening everything. I would recommend just specifying the post-grad stuff.

B

B

Where's, my no.

B

So specified ports- that's TCP, oh yeah,.

G

And then just five four three two also.

B

G

That's only in the secondary, but we'll we need we can. We can add it now and then we only need one firewall rule for okay.

G

And then, if we go back to the.

G

Pm instance, we should be, will apply. This tag, I think yeah.

B

I was I was wondering how I, maybe I, should have done sort of set a source tag here.

B

Like applicable,.

B

Compute engine, it's called DC cube.

B

Why isn't there a TC? Oh.

B

I'm getting up into a wall here for some reason: how did that happen.

F

B

What is cloud console is also being slow today.

E

F

E

F

Sing I was just curious where you're located I'm from Belgium Belgium nice.

B

Or these automatically created are these or these before yeah, of course, era for DES yeah.

G

There are four different clusters.

B

Okay: let's go ahead again: DC q, DB locks off Network yeah, which network artists on network default, allows specific targets. Ip range.

B

Q database no specific ports.

B

So these are sort stacks now. I think that.

B

Okay and now we can cube, can we maybe refresh.

B

Yeah I have to drop open, often a little bit, because I have to pick up the kids I'm, pretty sure we won't make it to the end of the demo in a short about a century.

A

B

It's you know connecting.

A

You know the demo is not having a problem because of the actual the environment setup.

F

Sorry, Steven, you were saying that was Jason. Oh.

B

Yeah I saw you both talking I thought, but I noise.

D

Yeah, these cans at demos are super valuable right because it helps us understand where the rough spots and the sharp edges are.

B

D

Ashley there were.

I

B

People more and like customer, ending in sales and stuff like that who wanted to join his demo I was like well: let's not do this for this demo, because I'm not sure it will be very smooth, understood.

B

Did this work now I'm, not sure.

B

I'm not sure what I'm doing at the moment.

B

This is a confusing part for me.

B

So I have grouped you: I have white type ingress. What does it mean.

B

Okay, please see what is it again cube demo.

B

It's applicable to instances.

B

Yeah I'm back at the machine.

B

Tacks cube, TV and I also have set it here. I think that that's the thing to do normally or I can.

B

Service, no, that's what I.

B

Target all instances on the network source, filter, I.

G

Just started responding for me on.

B

This one uh-huh.

C

G

Right so now we have a database connection.

G

We have a database connection right so now the migrations need to run. So we can go sorry.

B

I can log accept the task, runner I, think.

G

So yeah and we can do the logs against the migration pod.

B

B

B

8:28 minutes ago so.

G

They should start if we do logs on this. We should see it running the actual.

B

Alright, actually I have forced to follow. Alright, that's missing towards me. Trying not.

G

Seeing database is not an admission last year, so that means we need to operate the update. Oh there we go yeah.

B

So it's working with internal I piece. Now, that's what you weren't wanted to say that I guess I was.

G

Saying something different I didn't expect it to work with internal IPS, but I'm happy that it is I thought we needed to update to connect to the external IP, but no we're good.

B

Yeah we said set, the firewall will pretty wide, so yeah.

G

I thought they kept their their internal IP is like completely isolated like. If you even wanted to talk between networks, you had to use the external IP, but either I'm wrong or where we put the cluster on the same network as the instances.

B

B

What is this saying.

B

That was a little bit too ambitious. I guess.

G

Well, we can do a cute control, get pods.

G

So we're still, you should see the unicorn pods only have one of two running, so we're still waiting for those to finish starting up.

G

If you do it, if you do the queue control get pots again, but adding a dash W.

G

That'll update them updates in real time, as the states change we have. We do have two of two unicorns now, so if you want to try real oppression again, we might be able to.

G

B

Yeah I've seen this password somewhere fly. Oh yeah.

G

So now, if you do the right there.

G

B

B

Well, this is pretty cool already.

B

At least from my point of view now, I can finally add the license yep.

B

That was just a little minor step and.

B

B

B

If you don't mind, I'm gonna stop here, because I really need to leave. That's fine with me. Yeah.

G

B

Or did I miss is a new standard from Jason which piece yeah. He was saying something that the next part was was changed recently.

G

Maybe we did just add so the now the we're not sure has actually merged yet, but it hasn't been released for sure that running the geo migrations on the secondary once that's deployed so in here it should be a manual step on the server and that's no longer required uh-huh, but other than that everything should be the same. You shouldn't need to run the migrations on the secondary I can.

D

Ask Jason to follow up as well.

G

Yeah, if you just want a thing in slack, if you're working on it later on I'll, be around all day, yeah.

B

I really appreciate for being here, helping you I, wasn't sure what was going on so I, really no.

G

Problem, it's a good spot reverse to see where the pain points are so yeah.

B

I'm not sure what next steps are from here. Yeah upload recording might be interesting to anyone who wants to see it. Maybe he'll try to or or will dude extract notes from what was wrong or how will we do that? I.

G

Guess maybe we both want to do it cuz they both have different viewpoints as to okay.

B

Yeah yeah, that's that's a great point. I agree: yeah I started talking.

D

And people can just throw feedback in there. Oh yeah that.

G

Sounds good yeah, I.

B

Agree: well thanks. Everyone for joining we're only tween left now what okay.

F

D

Folks sitting down- and you know we're breaking new ground, and so folks, you know getting together and then making it happen is hard to get lab way right. It.

B

Was exciting for me as well, so that was very useful. Thank you to see.

F

Around I guess.