Magento Cloud Community, 20 Feb 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Magento Cloud Demo - Cron and locks + warm-up with Bohdan Korablov and Oleksandr Shmyheliuk

Description

Bohdan will tell you about the architecture of Magento cron functionality and how it's working on the Magento Cloud infrastructure. Oleksandr will cover Warm-Up functionality and implementation.

A

My name is Shankar blue and today I would like to talk about chrome improvements and some fixes. So, as you know, we release the TC tools: 200 200, oh and magenta cloud, watches 101. So these packages contains some improvements and features for Chrome. We fix it problem with skilled jobs, because various implementation can run, kill it or diet jobs.

A

Only in 24 hours and previous implementation had some problem. When some job takes a long time. This job can stack all jobs from the same growth, though it may fix it, and a little reminder about run schedule table this table contains scheduled jobs and has the next statuses painting it's when jobs waiting for their time to run running status when jobs in are in progress, miss it is where jobs didn't run because a scheduled time timeout and error it's when jobs through some exceptions or another, had another errors.

A

So it's all. Implementation is high levels him. This him shows how chrome work it before when you round common chronograph. This comment tried to walk group. If work was secured, we clean up whole schedule and generate new schedule of this group.

A

If this group was locked before so, we cannot acquire acquire lock now we do nothing and stop protests for this group.

A

So if group lock it, we try to run some cron jobs. We spending status from this group and we try to set around running status in schedule table for this job. If we can set this status, we set set status and check time scheduled time for this job is okay. We check it configuration for this job like configure it method on instance, type. If job has some problem, we set our status and and process for the job and group.

A

If it's okay, we're on job, if job didn't strong exception, we set a success status and stop it process if job throw with some exceptions. We save message from this exception and set error status. Is this part we had some problem like if job it's long time, we looked all jobs from the same group, new implementation? We change it to this behavior in this part and for loops. We currently we use magenta frame or logs, so we lock it groups on now. We have got it rules only for cleaning schedule.

A

With the country we don't use running status for working job, so if currently, if some job kill it by user or operation system and now schedule table contains running status and we next time run command. Crong crong me check it work for this the job instead of running status. If we can acquire this lock, we set error status for prayers, drop future and set running status for current job and next part like without any change.

A

What about this fix? It is this fixes this fixes will be released in Magento 2.3.5, but we created patches from much much for magenta from to the 2.52.

A

2.3.4- and maybe you know we have visitors- comment cron unlock. We use it. This comment when some job was kill it and we use this common to change running status. But now this comment we can use only part of previous much intervention if much intervention less than 2.5 to 2.

A

And currently, as I stayed if some job takes a long long time, this job doesn't stock all jobs from the same room and few walls about logs on the cloud we used to providers from working DB and file walk, but magenta support, so t prologues to back on cloud. We use only DB and file log on staging or integration and start environment.

A

We use DB loss because this environment have only one mode for database and this law works fine, but on pro environments we have C database a lot and sometimes when fire water works and change connection, you can lose this loss. So, by default on pro environments on staging and production, we use file log.

A

We have special file system. This file system is like proxy for zookeeper servers.

A

This is right on the file system and usually, if work, everything works. Fine, you would not see any files on his bus pass for mounted directory with this file system on the environment. Magenta cloud works dear. This variable contains path for this delivery. Micron project ID box. If you want you can change these web providers. For example, you can use database log, but we don't recommend use database, welcome pro environments on staging and production by default, easy to set up provider files on pro environments and PC.

A

Tools gets passed from magenta cloud box variable and sent to a B TCM PHP file during ready plan.

B

Hi guys, my name is Lucinda schmuck I'm from cloud Tim and today, as you may know, cash warm-up was added in previous releases, but in the latest release we edit some improvements and new features sites such as product warm-up product page warm up, store, page warm up for multiple stores and another.

B

Why do we need to warm up cash, easy tools after each deployment cleans cash to avoid situation when old cash data loaded for new functionality, without cash, which loads much slower and as a results you can? You can lose some visitors because, according to statistics, 40% of visitors will left the site if page loads, more than four second to reduce possibility of such situation, we recommend to warm up cake cash exactly after deployment phase.

B

Cache warm-up doesn't effect that one time at all, because it's running on post-deploy, face and site is already live and it's running around.

B

The which page should be warm-up it, it depends on your project. All project is unique. You need to find some may be more popular pages like popular products, categories, etc. It's all depends on your project, but we recommend to few popular, be popular products or category.

B

Of cash cash warm-up works, these heaters collect a list or else according to given configuration and after that it sends get requests in chronic related to collected pages.

B

Stop to configure have to configure cache for configuration for my pages, you need to add variable. Warm-Up pages into post, apply section in magenta and VML file. Here you can see, there are all examples of configuration types. We will analyze them on the next slide, so you can figure action types. Do we have? It can be single page, this full URL, it relative players and it can be multiple pages, and for this we're using some kind of pattern which consists of three parts. The first part is entity type.

B

It can be category, it's a math, page product or custom. Url second part depends on entity type. It can be reject pattern, part of URL of product, sq and center-set part a it can be a story, I do store code or you can use aesthetics on stores.

B

Let's see some examples for each configuration type.

B

Just configuration tag it's a relative URL; it will be combined it with well of your default store. For example, if you have only one storage, it can can work for you example, if you put on the slash it will warm up your contact hot deals. It will combine with your base. Url also you can pull, but in this case, what should be registered as one as URL of one of your stores. In other case, we do not remove this page.

B

Also, you can warm up single page for multiple stores release so, for example, if you of several Taurus with same structure.

B

You should use this pattern. Just parties, entity type called store page, then the patient. What you want to warm up and third part. It's store idea of store cause if you put Asterix it'll warm up, Contact, Us page for all the stores. In next example, you can see how to warm up by age, four stories, ID 1 and in the last one you can see how to come up each cars, full stories with code store in and storage EBM and.

B

Next configuration type is guest category and same as page if they work, but it is the same but difference on the in first part of the pattern for category its category and fitzsimons page its same as page and the second part can be reject spatter. So we collect all category rules and much to this pattern and also it can be just.

B

And the last part its previous to register code. So here you can see several examples in the first examples: easy to school, warm up all category pages for all stores, but use this process, because it can be very big amount of categories. It's also, it's always depends on project second example: government, all category or categories for two stores with store code store, F, are in stores D in the next example. You will warm up. We will room up page category page cup for the Swiss store in.

B

If such category pipe page exists, if not exist, we will not warm up and write appropriate look into cloud lock. File- and here you can see- is the same example for the CMS page and the last example. How to use reject reject pattern. For example, you put most part category and then you second parties reject buttons. So it will warm up all category pages which contains tires and.

B

And last type of configuration its product pages, it's called it also was added in the latest release. So to configure to warm up product pages. You should use next pattern. First part its entity type called product like on the part its product, sq can be one or several excuse divided by five, and the last button is the same as for all previous configuration types, it can be stores organs.

B

Let's check some examples.

B

In first example, they will warm up all products, but we programmatically limited it to 100 products to avoid some performance issue because it can be big and if it multiply on the multiply to count stores can be huge.

B

Second example: we warm up, all products for store, is caught, store, a fire and but it also limited 100 products. So if you use products instead of product desk, you use aesthetics, it always will be limited to 100 products.

B

Next example, in the next example, you can see have to warm up product with sq1 for all stores. If such products exist in this store next example, you see have to warm up product. This is q1 for store with squats to refer and store them, and in next two examples you can see how to warm up several products which / pipes in this example. You warm up product, whiskey, 1, 2, 3, 4, all stores, and in next example you we will warm up project s Q.

B

This is Q one two and three four two stores this quote store a file and store them.

B

All information about remapping we write to cloud look such as how many URLs we warm up. If we cannot warm up some pages according to rules, you also go see appropriate matches and also you see whole list of pages photos which was our Muppet more more examples and information. You can find blocked.

B

That's all any questions.

C

There are few questions in chat, I guess I chose either for you. So for CMS pages contact is identifier and for category is that URL key.

B

I, don't fully understand this equations here you can see example 4 category and same as pages who category entity type. It will be category, and so CMS page will be same as page. But second part is the similar is the same for both types. It can be rejects pattern on our URL or the no difference between category SMS page configuration types only in entity type name.

B

And I see another questions about if sq is limited to 100 what what rule is used to select the 100.

B

So in case, when you use aesthetics for product, excuse will limit it to 100 products and only in case when you use aesthetics for the description. In other cases, we will not reduce amount of pages to warm up.

D

E

That the question there is is: how do we figure out, which 100 products do we use? Is it.

B

E

Id, what's the little order, I.

B

Think it will be by by order in the table just first stop. First, toe active products.

B

And next questions would be nice feature if you could warm up the set mod. Actually we can. We can warm up the set map you can just put relative or full pass for two sitemap and it will be warm up.

B

Next questions about how Pakistan get a request, video post for sending to get requests we using classless gtp library and to be sending this request in as a synchronous in parallel. So at one moment it'll warm up, I, don't remember exactly neighbor number like 80 pages time, something like that, but you will retreat from sitemap and no it will. It will not read from sitemap. We didn't add this feature.

B

You can add future requests on our github and we will. We will review it and.

B

At the moment, you don't have functionality to warm up all pages from sitemap.

B

This presentation will be uploaded to YouTube and we'll send link.

D

To YouTube channel, and then we process here else to the cloud Channel in the slack, and we also update the demo schedule page that we posted recently, so we are always free to go and watch this presentation. Thank you.

B

Do you have any more questions.

B

C

Thank you. Everyone by journeying and sang wannberg done for for covering these topics so looks like next. One will happen on March, fiefs and we'll be talking about the latest ecl's release cells. The topic will be actual, is a release demo for latest EC tools, so that one should be interesting, because you know this is one of the biggest releases which was being prepared for more than four months. So it is pretty much packed with new features.

C

If you didn't have a chance to try it yet definitely come on March thieves to actually join us for the other session will be kevin and then yeah always feel free to to contact us on slack. If you need any changes, I would say in the tool in itself feel free to post tickets on github. All the repositories are open and we're always happy to to review feature requests and improvement. Requests thanks. Everyone and see you in slack and on the next on the next presentation, we'll 7 March have a great day.

B