I'm working on a facebook chat bot, and I'm developing it using the serverless framework (Node.js) and deploying it to aws lambda. For the first few weeks, I just ran a local serverless lambda simulator using the serverless offline plugin and everything was working great. Yesterday, I finally decided to deploy it to AWS lambda, and now I see a significant drop in performance and consistency. Sometimes the bot takes 10 seconds to respond and sometimes it is instantaneous. The weird part is, on the lambda cloud logs, it always says the function completes in around 150 ms, which seems super fast, but the facebook bot simply doesn't mirror that speed. I am hitting a database, but the queries are definitely not taking anywhere near 10 seconds to run.
UPDATE:
I decided to try to test the bot my manually sending requests to the API endpoint using postman (which is basically curl). Every time the api responded instantly, even when I send the exact same request body that the messenger does. So it seems like the request is just taking a long time to reach the lambda api, but when it gets there it runs as it should. Any ideas of how to fix this?
If the API is responding quickly to your curl request, then the problem isn't on AWS end. Try matching when you send your request via Facebook to your app and when your app recieves it.
If it's getting held up on Facebooks end, Im afraid there isnt much you can do to solve it.
Another issue could be the datacenter your lambda is running in versus where facebook is. For example, using chkutil.com, you can see facebook.com seems particularly slow from the Asia-Pacific datacenters.
As it turns out, Facebook was experiencing DNS issues and has since remedied the issue.
Related
Right now I have a Discord bot that is in approximately 575 servers, and on the website I made it lists the current server count of the bot. Right now, my method is to log the bot in every 5 minutes on the express app for the webpage and save the current server count to be served to the client. This causes memory usage spikes whenever I have to log in though, and using a whole discord.js application for one function seems inefficient.
I tried using the Discord API endpoint, but that was extremely laggy because there is only an endpoint for listing all the servers, not just the count. The endpoint also can only send info on 100 servers at a time, so I'd have to make a lot of different requests.
I'm hoping that there's a way to do this that would use less memory but still be fast. I tried looking into discord.js's source code to see if I could just isolate the functionality I needed, but I wasn't able to even find where in the code the data is requested from Discord. If anyone is able to figure how I could do this, it would be greatly appreciated.
You can try using free online database as a way to "communicate" data between your bot and your express app.
For example, you can use Cloud Firestore. Every 15 minutes (or whatever frequency you want) you can have your bot save server count information (and update time too if you want) into Cloud Firestore. Every time a client loads up your webpage, it'll retrieve the data from Cloud Firestore and be able to display server count and last updated time. (Alternatively you could have your express app retrieve that data every 15 minutes and cache it to send to the client)
You can use this method to share other data from your bot to your express app too.
The solution I ended up needing was a Discord websocket connection. That keeps everything updated live without having to deal with the memory and caching issues that come with discord.js. I've had a few other questions after this one on that topic, check those out if you want to see more on Discord websocket connections.
I'm able to successfully load test my bot server by getting the proper auth token from Microsofts auth URL (basically through this page)
I was wondering if this was a valid test on the service considering that we're not actually hitting the bot frameworks endpoint (which has rate limiting)
Is there another way to load test a bot service wherein i can replicate the bot frameworks throttling/rate limits?
I ended up with using load test with Visual Studio and Visual Studio Team Services.
The reason why I used this approach is that you can setup full path of load tests. Azure Bot Service can be either Web App or Function App with endpoint prepared for receiving messages - using HTTP POST so in the end is just web service.
You can setup load tests for different endpoints including number of hits to selected endpoint. In case of Bots you can for instance setup test with 100 fake messages sent to the bot to see the performance.
You can read more under these two links below:
Load test your app in the cloud using Visual Studio and VSTS
Quickstart: Create a load test project
Unfortunately as stated in the documentation you linked, the rates are not publicly available due to how often they are adjusted.
Regarding user-side throttling- this should not actually have an effect either way as long as you simulate reasonable traffic, but even if you go a bit overboard, an individual user hitting rate-limiting would be functionally equivalent to just having a bit more traffic. The single user sending more messages to the bot is the same as three users sending the same amount of messages slightly slower and there's no limit for your bot in terms of how many customers you might have. That said, a user getting a message, reading it, and typing up a response should not put themselves into a situation where they are rate-limited.
However, regarding bot side throttling it is useful to know if your bot is sending messages too fast for the system. If you are only ever replying directly to messages from users, this will not be an issue, as the system is built with replying to each user message in mind. The only area you might run into trouble is if you are sending additional (or unsolicited) messages, however even here as long as you are within reasonable limits you should be OK. (i.e. if you aren't sending several messages back to a user as fast as possible for each message they send you, you will probably not have problems.) You can set a threshold for bot replies within your channel at some reasonable-sounding limit to test this.
If you would like to see how your bot responds in cases where throttling is occurring (and not necessarily forcing it into tripping the throttling threshold), consider setting your custom channel to send 429 errors to your bot every so often so that it has to retry sending the message.
I'm developing a chat-bot using api.ai for NLP and i'm stucked where i need to query some data from a database. And also i need to do some processing of those data. I'm not much clear how can I do this using the webhook. Any thoughts and tips would be really appreciated.
You might want to check out API.AI's getting started guide, the last part of which involves the basics of adding fulfillment to your agent which pulls from this getting started sample
We use Cloud Functions for Firebase in the sample but you can't call non-Google APIs or URLs without setting up billing. You can also use any other hosting providers (Google's AppEngine, Compute Engine, Amazon's Lambda, EC2, Heroku, Azure, etc.) for fulfilling API.AI agents, as long as you have a server that can respond to a HTTPS request, it'll work with API.AI's webhook fulfillment. To get started you can even use your local machine and tools like ngrok to tunnel the connection from a HTTPS URL to your local development machine: https://ngrok.com/. If your just starting out I'd recommend doing whatever your most comfortable with.
Am currently using Node-RED to process the data received through the webhook. The webhook receives the data in JSON and if you’ve setup your intents and entities correctly, you’ll get data that you can act on.
This can be achieved as a stand-alone nodeJS app that you host somewhere, like Heroku or OpenShift. But important point is that you need a server running a program somewhere to process the data that api.ai sends.
Am using firebase as my database and is integrated into my Node-RED setup and now I can do sophisticated queries like “how much did I walk last Wednesday”
Hi I've been struggling with this issue for a few days now. I have a simple node.js app that connects to Twitter's streaming API and tracks a few terms. As a term is found the client side gets a websocket notification. I've made sure that my OAuth credentials are only used by this app and that the connection to the streaming API occurs only on app start up. What keeps happening is I get a 200 ok response but the stream then disconnects. I have it set to reconnect in 30 seconds but it's becoming ridiculous. It seems to be fine for a few minutes after restarting the app and then goes back to repeatedly disconnecting. The error is {"disconnect":{"code":7,"stream_name":"XXXXX-statuses158325","reason":"admin logout"}}. I have ran the same app locally with multiple client connections and not had a problem. I looked into other hosting services but I can't find one that supports websockets without having to revert to a slow long polling option on socket.io (which won't work for my app's purposes).
Any ideas for why this keeps happening?
that error means that you're connecting again with the same credentials (https://dev.twitter.com/discussions/11251).
One cause might be running more than 1 drone.
If this doesn't help, join us on http://webchat.jit.su and we'll do our best to help you :D
-yawnt
With the free account on Heroku and the free account on MongoLab (not with the Heroku plug-in) I get response time of ~1000ms per request (single user, it is just me still, relevant to all requests, not only first one after a long idle time).
I've checked from my own computer + the same free MongoLab account and I get ~168ms per the same type of requests.
While it is still high, I want to ask regarding Heroku. Is it reasonable to have such poor response time, even with the free account.
Will the response time go significantly better when I pay them?
Mongoose, MongoDB (node.js) Native Driver.
Do you have any idea for me what to check?
MongoLab helped me realize my database is defined in Europe, while the server (Heroku) is in the US. They also told me how to clone the existing database to a new one in us-east-1 (from their web console).