Slack lambda acts as a bridge between Client Slack and TC Central lambda
Follow instructions in Deployment Guide
Follow instructions in Verification Guide
REST API's are documented in swagger.yaml
- Table
projectshas the following fields,
id (primary_key): Unique id for each project
description: Project description
requester: The name of the client user who initiates project request
createdAt: Timestamp at which the project was created
status: Current status of the project. It is one of LAUNCHED, RESPONDED, ACCEPTED, DECLINED or APPROVED.
Why we need it: To handle scenarios like multiple clicks on buttons.
clientSlackThread (has an index - client_slack_thread_index): The thread id in client slack where the @topbot request command was initially invoked.
clientSlackChannel: The channel in client slack where the @topbot request command was initially invoked. The client can invoked this command in any channel so the combination of clientSlackThread and clientSlackChannel uniquely identify a request thread.
Why we need it: clientSlackThread, clientSlackChannel: These fields are used by Slack lambda to post a response to the right thread when /approve is called on it. It is also used to identify the project when an email command is issued. Without these fields we'd have no context of which thread to post responses to.
slackTeam: The team id of client slack workspace
Why we need it: To obtain the bot token used to communicate with client slack from the slack_clients table for this team.
tcSlackThread: Thread id in topcoder slack where the project request was initially posted by TC Central.
Why we need it: This is used to post messages to TC Slack when routes /accept or /decline or /invite is called. Also used during dialog submission. Without it we'd have no context of which thread to post responses to.
There is no need to store tcSlackChannel as it is fixed.
teamsConversationId: Conversation id in MS Teams where the @topbot request command was initially invoked. Conversation id is a combination of channel id and message id so we don't need an additional teams field to store channel information (like clientSlackChannel)
Why we need it: This field is used by Teams lambda to identify the project when an email command is issued. Without this field we'd have no way to identify which conversation to post responses to.
platform: Supported messageing platforms. Currently it can be either slack or teams. It is used to differentiate requests from Client Slack and Client MS Teams.
- Table
slack_clientshas the following fields:
teamId: The id of the team. This is the primary key used to identify the team in which the bot is installed.
userId: User id of the user who installed the bot into the team
userToken: Encrypted access token of the user
botToken: Encrypted access token of the bot. It is used to post messages to client slack.
You might see duplicate acknowledgement messages in Slack when issuing email invites.
The reason is that Slack expects a response within 3 seconds but since we're using lambda we can only return from a function after all our operations complete
So if we don't respond within 3 seconds, then Slack posts a duplicate event for the same command.
From docs,
"Your app should respond to the event request with an HTTP 2xx within three seconds. If it does not, we'll consider the event delivery attempt failed. After a failure, we'll retry three times, backing off exponentially."
The solution seems to be to have one lambda function call another using something like AWS SNS
See https://stackoverflow.com/questions/31714788/can-an-aws-lambda-function-call-another