POSable Documentation of App Tools

Introduction This document provides the technical details about the POSable app. It should be used to support continuing app development.
The POSable app uses many tools to support development and functionality. Development Tools include GitHub, Sprintly, Webstorm, Slack, Postman, RabbitMQ management dashboard, Loader IO, Circle CI, and New Relic. Tools used to support the functionality of the app include Node.js, Express, Amazon Web Services (AWS), RabbitMQ, and Mongo DB. Some of these functional tools are Node Package Manager (npm) items, however, not all the POSable npm items will be listed or discussed here. Also shown here will be app configuration details

App Configuration Details

Env Json File "mongoose_connection", persist http request data "mongoose_config_connection", persist user configuration data "logging_type", Bunyan setting "logging_level", Bunyan setting "logging_path", path for log files (needs to exist outside the app directory - will crash app if missing) "logging_period", Bunyan setting (one day). "redis_connection", connection string for redis database "wascally_connection_parameters" : { "user", user name "pass", password "server", amazon box "replyQueue", "port", "vhost", "level1_retries", see documentation on retry functionality "level2_retries",, see documentation on retry functionality pos-api/.ebextensions Two scripts are stored in this directory, 1. Install github key config and 2. Setup a logger directory outside of the app directory.

Config Records This shows and describes the fields saved in the config database for Merchants. name:, name of merchant posapiToken, token provided by POSable to identify user posUsername, point of sale username provided by agregator posPassword, point of sale password provided by agregator activeStatus, timezone, provided by agregator posVendorID, provided by agregator internalID, posable unique id for a user responseType, Alt will send an e-mail as well as regular response for http post related error messages. merchantID, provided by agregator accountingIsReady, boolean toggle that allows account data to be issued to accounting software, e.g , quickbooks invoiceConfig: visaID, provided by merchant mastercardID, provided by merchant amexID, provided by merchant discoverID, provided by merchant cashID, provided by merchant checkID, provided by merchant creditAccountID, provided by merchant cashAccountID, provided by merchant checkAccountID, provided by merchant accountingCustomerID, provided by merchant batchHour, provided by merchant batchMin, provided by merchant accountingClient, e.g., quickbooks cloudElemAPIKey, provided by cloud elements Organization, Element, salesLineItemID, taxLineItemID, giftLineItemID, discountLineItemID, batchType

GitHub The account name is POSable. In this account the primary (important) repositories are posable-api, posable-validationPlugin, posable-customerConfigPlugin, posable-wascally-wrapper, and posable-logPlugin All the plugins listed above including the wascally wrapper are available to each service as a custom npm item. Plugins are loaded during service deployment and must maintain a version.

Sprintly The account name is POSable. Current practice is to complete sprints every two weeks starting on Wednesday.

WebStorm The POSable team uses the code editor Webstorm developed by JetBrains. Licensing is provided by POSable

Slack Slack is used for internal development communications. The account name is POSabe.io. Channel names and use are obvious.

Amazon Web Services (aws) Account# 788743476555 It is planned that all production services will be hosted on AWS. Bastion server setup: Stage environment setup: stage-bastion-c Uses ssh agent to connect to bastion and from there to instances security groups all open ssh traffic only to bastion server https://blogs.aws.amazon.com/security/post/Tx3N8GFK85UN1G6/Securely-connect-to-Linuxinstances-running-in-a-private-Amazon-VPC IP of computer accessing the bastion server must be approved and configured in aws. To log into the boxes the public key provided with a pem file must be copied into the authorized_keys file. This link gives direction to find the public key. https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-key-pairs.html#retrieving-the-public-key VPC: Default Only RabbitMQ server for testing environment Stage 3 public segments (avail zones a, b, c) 3 private segments (avail zones a, b, c) Route table stage-route-public routes out to internet gateway igw-c5aed8a0 Elastic Beanstalk: 5 applications Each load balanced, health determined by health check Backend servers need health to be determined by messaging Do this with custom counter showing time messages in queues VPC for each app across all three availability zones Auto scaling Going forward, need to create AMI images for speed starting up Need: Alarms configured Need: email addresses subscribed to events Code has scripts for Beanstalk to run Sets up directories Creates SSH key for pulling plugins from private repositories Mongo DB Server: stage-mongodb-c All db’s on one server Need: clustering in production connection string is private DNS RabbitMQ Servers: rabbitMQ (dev instance) stage-rabbitmq-a, -b, -c are stage instances, clustered. See RMQ doc for ports Load Balancers: Two custom LBs for stage rabbit box and the dev rabbit box All others elastic beanstalk LBs S3: posable-awe-deploy-keys - ssh key for posable repositories All others automatically created - beanstalk ElastiCache cluster stage-cache-cluster Redis engine Connection in config plugin Route53 CNAME record added to posableapi.io stage.posableapi.io -> pos-api LB DNS SNS Topic ‘TestStage’ set up for notification service All others Elastic Beanstalk notifications.

Postman These three links provide the http requests used to test the POSable endpoints for fullTransactions, transactions and getToken. They include “happy paths” and “error paths” for xml and json payloads. https://www.getpostman.com/collections/3d5211285e816d92f8b5 https://www.getpostman.com/collections/a635c1260460794cc4ef https://www.getpostman.com/collections/48b49bdf56e6bf2e5e65

RabbitMQ Physical Deployment: AWS hosted dev RabbitMQ server (RabbitMQ) - single instance, connected via Public DNS 3 Stage servers clustered (stage-rabbitmq-a, -b, -c) Load balancer for messaging TCP traffic (stage-rabbit-int) port 5672 -> 5672 Load balancer for management site (stage-rabbitmq-test) port 80 -> 15672 Security group opens ports 25672 for cluster operation Config connection is load balancer DNS General Setup: All exchanges marked as durable All messages sent as durable All queues marked as ‘ha’ which mirror messages to other servers in cluster ‘Commands’ = many logical requestors, one logical servicer. Command message type naming is: [targetService].command.[commandName] ‘Events’ = one logical publisher, many logical subscribers. Event message type naming is: [originatingService].event.[eventName] Pre-Fetch limit (throttle) set for consuming services. Exchanges: all-commands: direct exchange route to queue using routing key (key matches queue names, listed below) used for ‘commands’ (many services commanding one logical service) all commands thru this one exchange event.deadletter: direct exchange routes to dead letter queue internal.delayExchange: fanout exchange only subscriber now is delayQ queue All other ‘event’ exchanges: All fanout (pub-sub) One exchange per ‘event’ (one service raising event to many logical services, pub-sub) Allows granular event subscription Events all flow to subscribed queues Queues: internal.delayQ: used for 2nd level retries (see below workflow) no handlers in code service.deadLetter used as ‘poison message’ parking no handlers in code All other service queues: Named service.[serviceName] All commands keyed to this queue deposit here All events subscribed to by this service deposit here Internal retry procedure: Code will not just ‘nack’ a message, returning bad messages over & over to queue. If error is in message format, and not system error send immediately to dead letter If error is system error: First X (configurable) times: messages have property defining ‘1st level retry’ counter WascallyWrapper created new dupe message, increment counter return message to RabbitMQ queues Next X (configurable) times: messages have property defining ‘2nd level retry’ counter W.W. creates dupe message, incrementing counter Sets routing key to be queue for consuming service Sends message to ‘delay’ exchange, parks it into delayQ queue delayQ set with ‘TTL’ policy at end of this time, message follows ‘deadLetter’ policy, moves to dead letter exchange because of routing key, dead letter queue sends to proper queue for redelivery Next: when out of retries, WW moves message to dead letter exchange with dead letter key Policies: ha_&_deadLetter: applies to all queues except the delayQ ha-mode: ‘all’ - queue durability to clustered servers dead-letter-exchange & dead-letter-routing-key: the definition of where to move messages when dead lettered, and the routing key - generally unused here due to code managed retries level2Delay: only delayQ ha-mode: same as above message-ttl: the amount of time message spends in queue before dead-lettering dead-letter-exchange: specifies ‘all-commands’ exchange, and since there is not change in the routing key from original, moves message back to handling queue

Loader.io Account - [email protected] Password: posable Pos-api will be the only service hit with http requests. A token had been placed in the app directory to verify the use of the test on the app endpoints. The directions below explain this. https://support.loader.io/article/20-verifying-an-app Past test results are accessible. App performance when load tested increased dramatically by changing nginx settings, specifically “worker_processes” and “worker_connections”.

Circle.CI Account name Posable

New Relic Account name: POSable, LLC Username : [email protected] Password: Posable1215

Redis

Wascally ** Keep alive settings between Rabbit and AWS are important to ensure connections are not dropped. Memory leak identified when connections are constantly re-established - connection object not garbage collected. This is a poor side effect of Wascally. Bunyan

NGINX

Quik Books [email protected] Pos@able12