2024 06 10 18 30 03

Summary

Встреча команды, обсуждение решений проблем с платежами, включая создание тестовых случаев для отмены транзакций и возвратов средств. Рассмотрены варианты автоматизации процесса возврата средств, обсуждение обновления логики хранения транзакционных идентификаторов и статусов. Также затрагивается вопрос о необходимости обновления платформы для поддержки новых функций и обсуждение миграции инфраструктуры на Kubernetes без простоев.

Takeaways

• 🕒 Встреча началась с опоздания и некоторой заминки в начале.
• 🔍 Команда обсуждала решение проблемы с отклоненными платежами и создание тестовых случаев для обработки таких ситуаций.
• 📝 Необходимость внесения изменений в логику хранения и извлечения идентификаторов транзакций для учета отклоненных платежей.
• 🔧 Разработка требует дополнительного времени для обновления логики и реализации требований.
• 📧 Планируется отправка уведомлений по электронной почте клиентам в случае проблем с платежами.
• 🔄 Обсуждение о том, следует ли обновлять базу данных для хранения статуса транзакций, включая не только успешные, но и отклоненные.
• 🔒 Важность корректной работы с API-соединениями для автоматизации процессов, включая интеграцию с платежными системами и бухгалтерскими программами.
• 🛠️ Упоминается необходимость внесения 'горячих' исправлений и улучшений в пользовательский опыт после запуска продукта.
• 📈 Планируется баланс между разработкой новых функций и устранением ошибок для обеспечения высокого качества приложения.
• 🤔 Обсуждение о том, как обрабатывать частичные возвраты средств и отмены транзакций, которые не поддерживаются на данный момент.
• 🌐 Говорится о миграции с Docker Swarm на Kubernetes для инфраструктуры производства, с акцентом на необходимости тестирования в новой среде.

Q & A

Что является основной проблемой, которую команда пытается решить?

Основная проблема, с которой борется команда, заключается в обработке сбоев в платежах и создании тестовых случаев для отмененных транзакций.

Какие изменения в логике хранения транзакций предложил участник?

Участник предложил обновить логику для хранения идентификаторов транзакций независимо от их статуса, включая и отклоненные транзакции.

Что такое 'The Round Up For Everything' и как оно связано с проблемой?

'The Round Up For Everything' - это идея, предложенная одним из участников, которая, возможно, связана с обработкой возвратов средств в случае отклоненных транзакций.

Какие действия предполагает разработка тестовых случаев для отмененных платежей?

Разработка тестовых случаев предполагает запись идентификаторов транзакций и их статусов, таких как 'declined' или 'rejected', для последующей проверки и возврата средств.

Какой статус транзакций в настоящее время хранится в базе данных?

В базе данных в настоящее время хранится только статус успешных транзакций.

Какой вклад в разработку ожидается от участника丹尼尔?

Участник丹尼尔 должен внести изменения в логику кода для корректного хранения и получения идентификаторов транзакций, что займет примерно два дня.

Как часто проверяется веб-хук для успешных транзакций?

Веб-хуки проверяются немедленно после каждой успешной транзакции.

Какой срок ожидается для реализации изменений в обработке отмененных платежей?

Срок реализации изменений запланирован на конец дня среды.

Какие действия предусмотрены для обеспечения качества (UAT) в рамках проекта?

Планируется провести дополнительное тестирование и внедрение исправлений для обеспечения качества до запуска проекта.

Какие проблемы с доступностью возникли в ходе UAT, и как их решили?

В ходе UAT участницаКим遇到了 доступные проблемы, хотя для остальных проблем была найдена корректировка. Она вынуждена была отказаться от доступа, что затрудняет следование за процессом.

Что означает аббревиатура 'MVP' в контексте этого разговора?

MVP в данном контексте означает 'Минимальный Выпущенный Продукт', то есть базовый продукт с некоторыми основными функциями, который выпускается для получения обратной связи от пользователей.

Какие действия предпринимаются для обеспечения непрерывности работы сервиса при миграции на Kubernetes?

Для обеспечения непрерывности работы планируется использовать обратный прокси для обеспечения доступности сервиса во время миграции на Kubernetes.

Какие интеграции планируется использовать для автоматизации процесса реконструкции транзакций?

Планируется использовать интеграцию между платежной системой, программным обеспечением для доступа и учета в Zoho Books для автоматизации процесса реконструкции транзакций.

Что такое 'PCI DSS' и как оно упомянуто в контексте разговора?

PCI DSS означает 'Payment Card Industry Data Security Standard' - это стандарт безопасности данных, который упоминается в разговоре в контексте требований к хранению и обработке платежной информации.

Outlines

00:00

😀 Вступление и проблема с платежами

В этом параграфе ведущий приветствует аудиторию и объясняет, что они собираются решить проблему с платежами, которые были отклонены. Также говорит о необходимости максимально эффективного использования времени и просит участников немедленно перейти к обсуждению. Важной задачей является создание тестового случая для возврата средств в случае отклонения платежа.

05:05

🔍 Обсуждение предыдущих решений и предложения 'The Round Up For Everything'

В этом параграфе участники обсуждают предыдущие решения, касающиеся хранения идентификаторов транзакций и отправки писем клиентам при ошибках с платежами. Aron предложил 'The Round Up For Everything', что означает решение проблемы возврата средств в случае отклоненных транзакций. Также обсуждается реализация нескольких решений, включая обновление логики для хранения идентификаторов транзакций и обработки отклоненных платежей.

10:05

🛠️ Технические требования и разработка

В этом параграфе обсуждается техническая сторона реализации решения. Участник из разработческой команды сообщает о необходимости обновления логики для корректного хранения и получения идентификаторов транзакций. Также затрагивается вопрос о том, как быстро это может быть разработано и внедрено, а также обсуждается создание тестовых случаев и обработка отклоненных транзакций.

15:06

📅 Сроки и планы по разработке

В этом параграфе обсуждается временная шкала для реализации технических изменений. Участники согласуются на то, что изменения должны быть завершены к концу следующей недели. Также затрагивается тема тестирования и необходимость внесения изменений в работу с поддержкой клиентов и обеспечением доступности данных для тестирования.

20:09

🔄 Обсуждение текущих задач и обновлений

В этом параграфе участники обсуждают текущие задачи, включая обновление продукта, которое не было завершено в ожидаемый срок. Также затрагивается проблема с доступностью UAT и необходимость в добавлении функциональности для справочного центра в рамках платформы Access America.

25:13

🤔 Рассмотрение вопросов о возвратах и реконцилиации

Участники обсуждают процессы возврата и реконцилиации, включая автоматические и ручные методы. Обсуждается, как будут обрабатываться возвраты средств, если платеж был отклонен, и как будет выявляться и решаться несоответствие в отчетах о платежах и заказах.

30:14

💡 Предложения по улучшению и масштабированию

В этом параграфе обсуждается необходимость внедрения API-соединений для автоматизации процессов возврата и реконцилиации. Также затрагивается тема добавления новых функций и улучшений в будущем, с акцентом на обеспечении высокого уровня пользовательского опыта и корректности работы приложения.

35:18

🗓️ Заключение о готовности к выпуску и дальнейших действиях

В заключительном параграфе обсуждается готовность продукта к выпуску и дальнейшие действия по тестированию и уточнению деталей. Также участники обсуждают влияние миграции на инфраструктуру и как это может повлиять на производительность и доступность платформы.

Keywords

💡API

API, или Application Programming Interface, является набором инструкций и протоколов, используемых программами для взаимодействия с веб-сервисами или другими приложениями. В контексте видео, API является ключевым для обеспечения взаимодействия между системой Access America и платформой 40s для обработки транзакций и возвратов.

💡Тестовый случай

Тестовый случай - это специально разработанная последовательность действий, используемых для проверки функциональности системы. В видео, команда обсуждает создание тестовых случаев для обработки отклоненных платежей, что является важным аспектом обеспечения стабильности и надежности платежной системы.

💡Возврат средств

Возврат средств - это процесс возврата денежных средств клиенту в случае отказа транзакции. В видео, проблема возврата средств после отклоненных платежей является основной темой обсуждения, что демонстрирует важность обработки ошибок и обеспечения удовлетворительного опыта пользователя.

💡Отклоненная транзакция

Отклоненная транзакция - это ситуация, когда покупка или платеж не удается завершить по каким-либо причинам. В видео, отклоненные транзакции являются основной проблемой, которую команда пытается решить, обсуждая способы возврата средств и улучшения процесса.

💡Развертывание

Развертывание - это процесс установки и запуска приложения или системы в рабочей среде. В контексте видео, обсуждается готовность инфраструктуры для развертывания продукта, что является ключевым этапом перед запуском в продакшн.

💡Контрольный баланс

Контрольный баланс - это процесс проверки и согласования финансовых данных для обнаружения и устранения нестыковок. В видео, команда обсуждает необходимость автоматизации процесса контроля за балансами, что важно для корректного учета и предотвращения ошибок в финансовой отчетности.

💡Платежная система

Платежная система - это совокупность технологий и процессов, используемых для обработки платежей. В видео, платежная система 40s является центральным компонентом обсуждений, так как она обеспечивает функционирование онлайн-платежей и возвратов.

💡Манипуляции с базами данных

Манипуляции с базами данных включают в себя добавление, изменение, удаление и извлечение данных. В контексте видео, манипуляции с базами данных связаны с обновлением логики хранения и извлечения идентификаторов транзакций для улучшения процесса возврата средств.

💡Контрольная точка

Контрольная точка - это специальное место или условие, на котором система проверяет свое состояние для обнаружения ошибок. В видео, контрольная точка связана с проверкой статуса транзакций и обеспечением корректного взаимодействия с платежной системой.

💡Компоненты QA

QA, или Quality Assurance, обеспечивает контроль качества продукта или процесса. Компоненты QA в видео включают в себя тестирование и отладку функциональности системы, что важно для обнаружения и устранения ошибок перед запуском продукта.

💡Кластер Kubernetes

Кластер Kubernetes - это группа узлов, работающих вместе для развертывания и управления контейнеризированными приложениями. В видео, обсуждается миграция существующей инфраструктуры на Kubernetes для улучшения масштабируемости и надежности.

Highlights

Meeting begins with apologies for being late and an immediate focus on maximizing time.

A summary request for the current problem to solve: handling declined payments and creating a test case for refunds.

Discussion on the importance of capturing transaction IDs and updating the logic for storing them.

The proposal of 'Round Up For Everything' as a solution for refunding declined transactions.

Concerns about the development timeline and the need for quick updates to the logic for transaction IDs.

The plan to update the database to store all transactions, not just successful ones.

A question about the test case for recording database transactions and their statuses.

Ensuring the solution meets all requirements from Forest Pay based on their test cases.

Timeline estimation for the completion of the development, aiming for the end of Wednesday.

A potential issue with storing only approved transactions and the need to change this practice.

The need for hot fixes and improvements to the user experience post-launch.

Discussion on the readiness of the production environment and the migration to Kubernetes.

Concerns about the impact of switching from Docker Swarm to Kubernetes on production stability.

The strategy for a seamless migration to Kubernetes without causing downtime.

The importance of keeping track of transaction reconciliations for accounting purposes.

A debate on the feasibility of implementing partial refunds and their impact on order processing.

The need for a clear communication strategy with customers regarding product availability.

The consideration of post-MVP planning for features like partial refunds and their reconciliation.

Transcripts

again everyone

hello hello

hello hello

hello

oh for

sorry guys I'm couple

ofes all right I think we're waiting for

try I just messaged him for

oh I go back here

minut

[Applause]

hello everyone sorry I'm

late hello hey

hello all right we don't need to do the

purpose uh we did the purpose earlier

today and I want to maximize time as

much as possible um have did you guys

already start talking about things no

okay all right so I appreciate you guys

staying a little later to help us uh get

through this and see what we can get

done here um sir can somebody summarize

the by the way I've gone intense meeting

intense meeting intense meeting so I

need a I need a summary just to get me

back in the right head space here with

you guys what problem are we trying to

solve right

now anyone want to take

it uh okay I can uh go ahead so we try

to solve flow with declined payments I

suppose and uh make it

uh

like and create a test case for make uh

declined like refund for decline

transaction yeah so we we

also uh would like to summarize the

stuff that we discussed previously on

that uh stores a transaction ID uh stor

the status of that and send email to the

customer uh

if something went wrong with his payment

however Aron Al told about anything else

that would be a good uh experience for

e-commerce so Aron can you remind what

we are also trying to solve from your

perspective uh my uh my proposal is in

fact The Round Up For Everything uh so

we are trying to solve the problem for

refund for a uh decline transactions so

where it normally occurs the I mean the

purchase uh is failed or like a the

transaction is failed but still the

money got deducted uh from our customer

end point uh so how we can actually give

that money back uh from them uh the

reconcillation

part uh

yeah uh how we want to do it like

ideally the I propos multiple Solutions

one is like a

having in a complete degration or else

like a sending in a declared transaction

to 40s uh 40 pay

uh so that like they'll uh they'll

actually uh refund money to customer if

they deducted uh money they receive

money against those

transactions all right so what problems

do you see Danielle and

uh since we know now that we're going to

capture these transaction IDs what

problems you see in in getting this

developed

quickly uh so to get this uh you're mute

oh

yeah yeah sorry uh to get this flow in

the way which we require I suppose uh

from uh my side I requir to update

little bit the logic

for correct storing and getting this uh

transaction ID cuz I after our previous

call go to documentation which which

created Olaf and I put in chat just

screenshot regarding one uh

point that we can

U like get transaction ID and store it

in database after successful response

from 40's app

and uh it's way how it work now uh and

uh if we want to update for our purposes

with this decline transaction I required

to make updates to logic suppose around

two days from my side for this purpose

cuz currently we have uh like flow like

uh uh we have a transaction that went

through or or a and if it's uh went

through and it's successful everything

is okay if a transaction doesn't go

through uh wec message on to saying why

it didn't go

through

that's what I have for

now okay so from the development

perspective there are no issues to

implement that

it just requires like additional time uh

to do that so I think we're okay however

I'm wondering if for the test

case uh after we do this can we just

record this database transaction ID and

Status rejected like decl or rejected

whatever it shows there and uh pass it

through to the forest so this would be

the result of this test

case so uh Daniel uh Daniel one quick

question while we are actually

discussing it uh as of today we are uh

storing irrespec of sorry we are storing

only completed or like a approved status

successful yeah yeah right right so if

we just remove that check while we are

actually storing

it uh you mean remove status of uh

successful or not successful and store

everything store yes we need to right

yeah yeah yeah yeah yeah that's flow

which we will update with this logic

because currently we store only

successful

transaction yeah all we just need to

change I don't know like how it's code

is written but if there if we are

actually writing uh updating database

layer based on the status if just remove

that status check uh condition that's it

right uh yeah and uh I require to make

edits in this flow cuz previous this m

like code was made by another person and

I just

okay I'm just I'm just asking I'm just

asking and as we discussed we just need

to add on a send email but I mean that's

a kind of a good to have but I don't

know like either we agreed upon that or

not I remember but

yeah if that is a din stage sorry uh

Bine status we just need to send a

message to customer saying that your

transaction has been

declined yeah like it's how we discussed

previous on previous

call

yeah so are we taking action based on

initial based on your screenshot here um

on the process flow you know are we

taking action on the initial response or

the web hook response uh on web hook

response so we're waiting until the web

hook says it was successful yeah right

because before we actually complete the

order yes it it says that pretty much

okay just want to make sure okay

good um how often are we checking that

web

hook uh well I mean or how often does

that web hook run yeah I got it I will

check I not sure in what

time most often it's just immediate but

I just want to make sure we're not

looking at I think it's multiple delays

immediate after like

request for

yeah it looks like it works after every

successful uh

transaction

okay so we you get immediate responses

yes

good okay so uh I think uh we are all

okay with the solution

discussed will that meet all of the

requirements that Forest pay has if we

do that I mean I know they they ask for

a couple things that we just don't

do right

J yes

you're confident that we're get we're

getting everything we need with this

this is it this is the thing this is the

one last thing with Forest pay

correct based on their test cases yes

this is the one thing that we cannot

record previously and if we do this we

will record this and send this out to

them so we will cover all the test cases

that are sent by them and all the points

that they will be checking okay and

what's the

timeline uh I I think around two

days so the timeline is end of the day

Wednesday right yeah end of the day our

time so it will be beginning of

yours we can complete the rest of uat

before this

correct

yes how is that going

um yeah yeah AR do you want to answer

your from your

perspective I mean as for the today's

status we don't have any major critical

books from a

QSI we don't see it

okay and I haven't recoup I mean regroup

with them on end of the day

yeah I haven't seen the bo I have to see

the bo as

well

okay did you say

board you you see the board is that what

you

said the devop devop board oh the devop

board got it

yes well you've got um yeah that's fine

I was just thinking of the

I know we got some new new tools in the

in the pipeline here too so I wasn't

sure what what we're talking about as

far as boards but okay um so

Jim anything else that you can think of

from and how's uat going with Access

America from the US

participants they've been on there going

back and forth

uh I know Kim has had some problems

getting in from one of her

emails uh even though you fixed it for

everybody else it wasn't fixed for it

didn't seem to fix for her I've still

had horrible times getting into uat but

I've kind of given up on that which

makes it hard to follow the orderflow we

still need to add back in

the piece that we had there originally

uh for the help

desk so I think that was taken down uh

so

frustrated add back the piece oh for the

help desk within the axis America

platform is that what you're talking

about yeah the actual help

piece the pages that we're put there all

it is is content for us right from from

the business side it's just

content and whether

whether whether that content was going

to exist in foris pay or it's just going

to go into

rcms static Pages whatever we're doing I

can't I would love to say CMS and know

it was there but whatever I'm not GNA

argue the point I just we're just giving

them data and the same data needed to be

created anyway no matter where it was

going so so what's that status the help

stuff uh Kim and Mayo just went through

it they're they're

um they're making revisions to it based

on you know the site which is one of the

reasons why they need to understand the

site so we just went we just had a

meeting about that

um yeah that's about it

okay

um anything else anybody anybody need

anything else to feel comfortable that

we are moving in the right direction

getting everything done that needs to be

done I have a question regarding product

test today in the morning we didn't see

this um products that were expected to

be done on Saturday if I'm not

mistaken

uh so do you have any update on

that product

no got it okay and

any timeline for that no

okay I don't have any other questions so

uh na uh I remember like uh there's a

variant uh variations related uh issues

right right I I know like these are the

open bugs from Friday right uh what we

are

planning well it was um

Daniel I believe because you are very

busy today you did not upload it to U

right I didn't see any message or uh

yeah uh let me check

uh uh regarding Victoria's messages uh I

can update on this uh that you could

probably not receive the updates from

Victoria yet because she had a power out

page so she will complete it today

anyway but just a little bit late okay

so this so this answer it um yeah so the

answer is that it was supposed to end

the development was about to end this

morning and we were about to publish it

to QA for

testing at around at around noon but you

the what just Yulia said that we have

some uh technical issues that will be

that we will be solved tomorrow and you

will have this release and you have this

feature ready to be tested in uat

tomorrow I believe in the first half of

the day yeah so by the way like TR what

we are going to do with our

reconsiderations with the cancellation

as we said um it's also part wait the

cancellation is okay I meant

this wait I'm yeah

considering all the scenarios of

reconciliations not just happy

cases can you repeat your question again

please so the current use case of

another

reconcilation not I'm not sure following

you

sorry so declined refunds any other any

other refunds

which is missed from a 40 side those

kind of a reconation how we are going to

handle it are we going to handle it on a

manual basis if on a manual basis how

long uh because if order started flowing

in or are we able to scale

it well now what we did is not manual

it's automatic so I don't see any reason

why it's so it's scal I don't I don't

I'm not sure I understand the question

again because it's scalable and every

time you do a

cancel you will be not cancellations

reconciling Rec

caner

reconciliation

reconciliation okay

yeah so

it's I just read Because I had somebody

come to the door I just read the uh what

you were saying you were

asking how we're going to handle

reconciliations right now and it's a

it's a manual process so we're getting

daily reports and each day um Finance is

going to go through those reports make

sure they align with our

um so we had a report from our

deposits that Forest pay gives us and

then we get a report from our uh orders

on the access America side and they

should align if they do not align then

we're going to be looking

for uh to reconcile any any differences

and that includes triggering either

refunds or whatever needs to happen on

the side of of uh Forest pay is that

your understanding

Jim uh yes that's exactly my

understanding okay so we'll be doing

that until we

can um get the API connections between

the payment

processor and access America and

our

um Zoho books for bookkeeping software

which then should allow us to uh MCH

transactions in a pretty automated

way Jo books does have an does not have

an integration for Forest pay but they

do have an open API

that has detailed

documentation

right so yeah so we're going to put all

three of those systems together take the

report report from this and the report

from that and reconcile them in an

fairly automated way um somebody else is

going to have to keep eyes on that every

time there's a

difference they're going to have to call

that

out so right now that's Kim who's going

to handle that and

then we're going to be seeking an

additional hire

soon to help her because there hopefully

there will be a lot of transactions and

she won't be able to keep up with it

that would be great

right if she can do transaction

reconciliation in three minutes or less

we're in a lot of

trouble all right I don't think that's

going to be the case but uh a I lean on

you with the e-commerce experience that

you have if if we're missing something

we're not thinking about something

please let us know

yeah no it's not always ideal when

people hear manual but here we are we'll

get it done and we'll uh and denelle you

can count on some of the next stuff

we're going to want to do first off once

we launch that you should know

that we need people available for hot

fixes for working on any issues that

come up it it's going to be very

critical because it's going to be

customer facing any issues we have I

think understand and then and then uh

these other things that are going to

improve our efficiency are going to

be very high on the list of things to

work on next so we're going to have hot

fixes priorities hot fixes that new

features uh efficiencies new

features I don't want to be I know that

ideally we want to get to a place where

we're doing sprints every two weeks

based you know new new features and

capabilities and all of that but I want

to make sure that we're reserving enough

development

time and or higher so that we can

reserve enough de development time that

we are giving lots of attention to bugs

and uh user experience improvements not

just Feature Feature Feature that it

should really

be a well-tuned

um application that we're going after as

well

as giving customers more features that

make make their buying

experience better that make their

customer service experience

better

um why do I say it say it because I see

so many companies that they're willing

to just put up with so many bugs I mean

stuff that they've known is WR is wrong

for months or years even and they just

won't fix them because it's not new

feature or their team is overwhelmed or

whatever their case is I just want to

make sure we're not doing that if we can

help it at all um and and that speaks to

the size of the team as well as how

we deal with

velocity any

questions no questions from side I don't

want to waste your guys's time talking

about future I want to talk about now

what are we going to get done um you're

saying two

days so like this time in this time in

two days say noon in two days this will

be up and running and we'll be video

will be on it way yeah but also

regarding another test case probably

it's better to push to 40s another test

cases now CU if there are anys in

another test cases or any like for

example if 4 is required to uh record a

new one video of any uh test case we

will have more time time for doing this

until for example we wait now 2 days and

then we will

push uh all um cases cuz as I think

probably better to push it today and uh

we will get a partly result on it and we

will

have more updates from 40 site uh soon

until we will wait for so I'm going to

I'm going to C consider that but what

what

I'm what is more important to me sorry

some some something just going on in the

background there

um what's more important to me than

getting it out immediately and getting

some feedback is that we give them a

complete set of what they ask

for first and then if there's problems

we deal with those problems but I don't

want to say to

them you know oh we're we're continuing

to develop this or continuing to work on

that I don't want them to feel like

they've got days

because we're continuing to work I want

Jim to be able to say on noon on the

12th him to be able to call call for us

and say you got it now when when can we

expect it to to be approved when will we

get approved when will we get approved

and be on them but the only way we can

be on them is if we give them complete

data the first time around that's my

feeling Jim do you have an opinion on

this uh I've thought about it both ways

I don't know I

mean it does make us seem amateurish if

we go to them with less than that and

say we're still working on stuff but

they may also you know there's stuff

that

that I missed on the PCI DSS about you

know having to

keep you know full transactions of

everything that's a fail transactions so

I don't know if there's something else

they're going to Port point out I mean

there are the in the area where they act

asked for direct Integrations the

refunds the

voids we said we couldn't do refunds on

previous transaction IDs because we

didn't have that on access we also said

for partial refunds we're not doing that

on access so there are things that we're

going to we're not giving them

everything but we are saying hey there's

some functionality we don't

have so do we tell them that we're

adding this functionality right now and

go ahead and try and approve the rest uh

so what what I

would oh are are you you're is that

rhetorical no oh

okay B says that there's no such thing

as a rhetorical question that I feel

like I have to answer every question

asked of me so I just wanted to make

sure

so

um I think that it's worth a phone call

to them to say here's where we're at in

the

status of getting you this there's these

two situations that have come up that we

need clarification

on

um and that is voids and partial

refunds

um I'm kind of guessing they're thinking

partial refunds in case there was an

overcharge of some

type but I can't think of any use case

that we have and if all of those things

are working properly but what happens if

they don't work properly the only reason

I thought of that is because of we

mentioned

today this weird use case that I still

don't understand

fully and now I'm

going maybe there's things happening

that I don't

understand so is it worth having a

discussion with them to say we have no

you know correct as if we're wrong but

we have no

reason to provide a void

or a partial refund so we're either

doing full refunds or not at

all the other thing that might be worth

checking out as I think about it and I

don't want to spend too much time

thinking out loud with everybody on the

call and I do need to get to fabiio also

and ask him uh about production

environment and its readiness

um

but

uh oh yeah yeah

so partial

refunds is there a way for us

to within Forest pay if

somebody they they bought a product

and then there was something wrong with

the

price do we give them a full refund and

then recharge them again and do we have

a way of charging that without it

causing an

order um and this is probably on your

side Jim in the because I think it's

going to be a manual process it's not

going to be automated on on access side

at all yeah also if somebody receiv

received they damaged good and but they

were like well it's damaged but you know

it's still

usable sometimes client sometimes

they'll give them a partial

refund or a store credit or something

like that um they could have also gotten

the incorrect

item and they like oh I wanted the 12in

statue and I got the 6in statue so it's

the wrong price so that could be a

partial

refund yeah so well maybe not ideal they

have to return the product that's

damaged they have to return the product

that's not the right product and they

can order the new

product I'm not I don't we don't have to

do that other thing like partial refunds

and all that other in fact from a from a

business perspective is dangerous

partial refunds can be avoided there by

returning the product but what if like a

customer ordered you know product and

for say like we have an inventory of two

products but when we are went ahead and

Tred to pack it only one product is

actually usable either we need to cancel

the complete complete order or like we

we have to go a partial ORD partial

refund if if we connect to the customer

and say that hey I'm okay with a single

item and probably like we might need to

ship that item and like give the

remaining partial orders I mean like

that

yeah there will be other cases as well

like a you know per say like in a

variant how per say like I ordered a

black cap uh but uh I you have a gray

cap the price difference is like a 2

but customer still okay with a greater

all right I'm just giving an example

right so I'm just thinking can we can we

run those manually and it you know in

the background are we able to do that so

when I say you you can do it manually

but how do you tag those transactions

with a order because end of the day your

entire

accounting right will be mapped to that

until and unless you have an way

to you know every transaction should be

mapped with order if until and unless

you have that way it's not really

possible see TR like you know these are

I know like probably like hey we want to

have all these things but on a day one I

never see any of on e-commerce companies

have this even on even on Amazon level

they launch a new articles right Prime

now and fresh they don't have this

things on a day one and we can still uh

communicate to our customers say that

like hey we regret to inform that like

this product is not there why don't you

we'll we'll actually like uh cancel it

please go back and uh you know uh Place

another order but if at least like if we

have another channel to communicate to

them say this say that like hey this is

the thing people will be happy but if

you want to build all those things on a

day one trust me it's a next to

Impossible because there's in ecommer so

as I mentioned earlier right like we

have a thousand use cases in a

e-commerce and we and none of the

e-commerce companies are fully

implemented what they want to do it to

satisfy the customers well so I'm I'm

happy with what you're saying because I

was G just I was literally just about to

ask is it something we can do Post MVP

because I think we have to if if we can

and so um but the only reason I bring it

up at all is because we're about to go

back to them and say we have no ability

to give a

partial I mean we can unless unless we s

unless we take a video of us doing it in

the platform Jim like yeah we're gonna

do it in the platform this is how we

would do it in the platform

I don't know but that's something that I

think you should ask them is like is it

okay can we and is it okay if we do it

in the platform partial return but and

you should be able to put in the notes I

I haven't se I haven't seen the platform

in a while but you should be able

to track it to an order well it has a

transaction ID and if you're

giving you know part part of the money

back to and it's related to a

transaction ID we have to have a way of

distinguishing the line item that it's

being applied to

right

yes so if you think about it kind of

like an

invoice the different line items you

know line item three or this product ID

or whatever it is we just have to be

able to track it to the exact item

that's been refunded or reason it's been

find it whatever because I you're right

Aron that that part of the order is

damaged part of the order doesn't show

up part of the

order was a wrong product whatever that

that's going to

happen and I wasn't thinking about that

use case because it doesn't happen to me

often because I order one thing at a

time a thousand times a day I'm

kidding that's that's my family not me

all and partial by the way like if you

want to implement a partial refunds now

you need to regenerate an invoice and we

need to look at like a when we are

actually generating an invoice you need

to updated invoice and back old invoices

everything in

place so there not a one thing you want

to do it now because it end up you

actually opening

[Music]

it's not it's not a right time to open

that box

now agreed I think we we're going to

need to

brainstorm a bit on the business side

pretty much

now what that looks like to make sure

that we have records that fully

reconcile across all

systems so I understand the financials

that's easy enough but what happens when

you start doing

refunds and they were you know in the

last month not this month which may be

the last quarter which may be the last

year if it was done you know on January

2nd the refund was done on January 2nd

the order was made on December

15 and it's end a year

close what does that do how do we do

that Kim may know these answers right

she's been in the retail side of things

forever she may know those answers

immediately I do

not all right we'll have to brainstorm

those later

Fabio welcome to the

call yep I'm here we we hold a product

meeting pretty much every day

um we have product meeting

for typically it was it's m Monday and

Wednesday for both access America and

Battery careers pre-launch for Access

America we're doing one every

day um so going forward I would like you

in that meeting you'll get an invite for

going forward this meeting at this time

is an unusual one it was because we used

all of our time in our early meeting

talking about what we were just wrapping

up there but what

what I need to know from you and the

team needs to know from you is whether

we're ready from a

production infrastructure and deployment

perspective are we ready and if not what

do you need to get

ready yeah it is all ready and you know

the takos architecture is uh fully

tested on De and uat and I figured out

all the things happened before so uh I'm

sure the takam is table for

production okay and uh the messages that

I received from you earlier

today oh it is for kubernetes

yes I understand but we're going we're

going to deploy two kubernetes for

production yeah and

then uh I think we need more time to

test on the kubernetes environment

because uh we didn't uh any QA on the

kubernetes server so all right uh you

know what I just forgot something I'm

sorry Fabio I I sorry for cutting you

off and sorry because I forgot that we

do have a full doctor

swarm cluster right now three neres

right

right

okay I I'm thinking before I say

something

stupid so give me a minute because I got

a lot of stupid things to say right

now I don't want to say I don't need to

say

say if

we deploy or if we maintain the docker

swarm

cluster and then we want to switch to

the kubernetes

cluster what downtime would that cause

and you

know uh the migration to the kubernetes

uh affect only the backend side you know

the front end is deployed on cloud flare

and the back end is deployed on tacos

now but we are going to migrate into

kues

so a

uh the steps like this so first deploy

the

uh Access America production project on

the kubernetes with another uh backand

API so you know the API gway access Ms R

is using for uh de environment so we can

uh deployed the access America backend

on the kubernetes with using other URL

like API Gateway uh K cage XS MSS and

then uh once the kubet is is stable and

uh all the database is migrated the

kubernetes and then we can change the

url on the cloud FL in that way we don't

have any uh down on on the

subers give it another

way put a reverse proxy in front of both

clusters and deploy both deploy services

to both

clusters and then and then when the

migration is complete it's seamless

fail don't we have a load balance

sort of a thing yeah that's what I'm

talking about putting a load there's a

load balancer in front of each cluster

but there's not one in front of both

clusters yeah makes

sense and then we can