brewskey / spark-server Goto Github PK

If a webhook is created and the selected URL times out, the server hangs completely with the following error:

uncaughtException { message: 'connect ECONNREFUSED 67.161.118.40:3000', stack: 'Error: connect ECONNREFUSED 67.161.118.40:3000\n at Object.exports._errnoException (util.js:1022:11)\n at exports._exceptionWithHostPort (util.js:1045:20)\n at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1087:14)' } [nodemon] app crashed - waiting for file changes before starting...

So we've been doing OTA updates pretty routinely over the past few weeks, and all has been working great. Today, the OTA stopped working (times out) and I don't see a reason why that would have happened overnight. We haven't pulled the latest update for the db yet. I've restarted the server which didn't help. Also tried flashing an older version of the firmware that that did not work either.

Here's a snippet from the error logs

2017-05-21 21:32 +00:00: (node:23436) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 4): Error: Request timed out
2017-05-21 21:34 +00:00: This client has an exclusive lock { cache_key: '_3',
  deviceID: '25001c001547343231323438',
  messageName: undefined }
2017-05-21 21:34 +00:00: (node:23436) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 6): Error: Request timed out
2017-05-21 21:36 +00:00: This client has an exclusive lock { cache_key: '_5',
  deviceID: '25001c001547343231323438',
  messageName: undefined }
2017-05-21 21:36 +00:00: (node:23436) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 8): Error: Request timed out
2017-05-21 21:48 +00:00: This client has an exclusive lock { cache_key: '_2',
  deviceID: '25001c001547343231323438',
  messageName: undefined }
2017-05-21 21:48 +00:00: You have triggered an unhandledRejection, you may have forgotten to catch a Promise rejection:
Error: Request timed out
    at Timeout._onTimeout (/root/node_modules/spark-server/node_modules/spark-protocol/dist/clients/Device.js:494:28)
    at ontimeout (timers.js:365:14)
    at tryOnTimeout (timers.js:237:5)
    at Timer.listOnTimeout (timers.js:207:5)
2017-05-21 21:51 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 3): Error: Request timed out
2017-05-21 21:54 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 4): Error: Request timed out
2017-05-21 21:56 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 5): Error: Request timed out
2017-05-21 22:02 +00:00: This client has an exclusive lock { cache_key: '_3',
  deviceID: '25001c001547343231323438',
  messageName: 'PingAck' }
2017-05-21 22:02 +00:00: This client has an exclusive lock.
2017-05-21 22:03 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 6): Error: Request timed out
2017-05-21 22:05 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 7): Error: Request timed out
2017-05-21 22:07 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 8): Error: Request timed out
2017-05-21 22:09 +00:00: This client has an exclusive lock { cache_key: '_10',
  deviceID: '25001c001547343231323438',
  messageName: undefined }
2017-05-21 22:09 +00:00: (node:1340) UnhandledPromiseRejectionWarning: Unhandled promise rejection (rejection id: 10): Error: Request timed out

When you trigger a webhook request to another server, I want to log all data we can get about the request as well as the success/failure response.

We should rely on some setting which tells us if we want to console.log these requests. It should also use constructor injection like all the other repositories so we can override the implementation.

When we use this for our server, I will want to log all the requests to a file so that if any errors occur, I can basically do "playback" on the file and get my database in sync.

We have a webhook that looks like this:

{
    "event": "power",
    "url": "https://ariocloud.azurewebsites.net/v1/integrations/particle",
    "requestType": "POST",
    "query": {
      "apikey": "notarealkey"
    },
    "noDefaults": false,
    "rejectUnauthorized": true,
  "headers": {
    "Content-Type":"application/x-www-form-urlencoded"
    }
}

Payload:
"0,btn"
The server is getting a payload like:
{"0,btn": ""}

These should be in the DB so we can start implementing the OAuth controller.

I think we need a repository for this which should populate the DB with defaults -- authClients.json

I'm not 100% sure how we will want to implement that controller because it should have a higher level of permissions than just allowing any user to create the tokens.

Update this readme to use the correct particle API. Also include a link to set up a windows machine.

Please remove the openssl line because all it did is break things for me.

Please format the commands so they are always on their own line. There were a few inline commands that I read over.

Hi, and good job.
I don't really know Babel, can someone help me to debug this project?

do we need this?

:( I'm lazy and haven't finished this yet...

I'd like you to make a migration script which would either take the files on disk and move them into the tingo/mongo db.

Start with this - https://github.com/sergeyksv/tingodb
For multi-server environments we'll simply need to switch to MongoDB (same APIs) which makes things easier :)

I chose this as it seemed like the easiest way for people to scale up their systems without having to rewrite a ton of repositories.

Use clustering to improve speed. We will want to do this at the HTTP server and allow it through settings on the COAP server (spark-protocol). We'll need a singleton way of passing events so you'll need to update EventPublisher (or whatever it is) to send data between the forks.

This should be really easy to implement on spark-server since we're stateless - https://github.com/Flipboard/express-cluster

We can hold off on this one until after webhooks/OAuth/device_claims changes have been made but the readme is definitely out of date.

Webhook response errors aren't caught inside onNewWebhookEvent try-catch and throw right into process.on('uncaught exception')

add device keys to db migratioin

Hi, I'm trying to set up spark server over a secure connection. I tried changing the port from 8080 to 443 in the settings.js file but that seems to have broken everything - tried sending commands to https://ipaddress:443 (+various permutations) but no luck. I think I've set up the SSL certificates properly but am not completely sure.

particle subscribe fails when we'are trying to subscribe to eventNames with /.
We get 404 error from api.

particle subscribe foo/bar send request to http://localhost:8080/foo/bar but the controller handles only http://localhost:8080/:eventPrefix

If I disconnect a device within a few seconds of seeing _ensureWeHaveIntrospectionData error: Error: can't get describe data the server will take several minutes to register the device as offline. In the interim period, requests to read variables from the device time error out with a 400, but only after 15 seconds. This is with the "coreRequestTimeout" setting set to 9000 (Changing it to 9000 from the default 30000 made no difference in this case).

(Devices have 0.6.0 FW)

I'm having trouble migrating a webhook from the particle server to the spark server; getting a return with "data": "Bad Request". This is a GET request with a query string (I have another webhook with a POST request with form data that is working fine on spark server).

Originally I had put the query string the URL like:
"url": "http://api.com/endpoint?id={{id}}&value={{value}}"

I also tried
"url": "http://api.com/endpoint"
"qs": "id={{id}}&value={{value}}"

But am still getting a bad request error each time. Any idea?

I'm getting this error on every request.
{"error":"server_error","error_description":"Server error: accessTokenExpiresAt must be a Date instance"}

now ouathClients are hard coded in .json
we need to realize that endpoint: https://docs.particle.io/reference/api/#create-an-oauth-client
so, one more repo/controller etc.

I think updating express will make things much cleaner and allow us to use better loggers.

It might be a pain in the ass to do it but we can look at this branch to see what changes they had to make to add it - straccio@dfe7984

Hi again, when i try to get a var value by requesting /v1/devices/:deviceID/:varName/ the response is:

{ "result": "[object Object]" }

You guys are doing a great job!

I was trying to build using npm run build && node ./build/main.js and i got this error:

.../node_modules/spark-protocol/lib/server/DeviceServer_v2.js:485 var _ref5 = _asyncToGenerator(regeneratorRuntime.mark(function _callee6(eventName, data, deviceID) { ReferenceError: regeneratorRuntime is not defined

hi guys i was playing with events and i would like to share with you how they should work for me:

1. GET /v1/events should return all public events (from all devices) and private events from devices owned by me
2. GET /v1/devices/events should return all public and private events from my devices only
3. GET /v1/devices/:deviceId/events should return all public and private events for a specific device if the device is owned by me otherwise only the public events

Is that the way you are going to implement them?

In your version i have:

1. GET /v1/events return all public and private events for all devices
2. GET /v1/devices/events doesn't return special private events like status / flash
3. GET /v1/devices/:deviceId/events return only events from devices i own

When you call a function from the cli and the function doesn't exist it should fail.

The CLI is expecting a different response then what we are returning.. We need to check the CLI code and figure out what response they are expecting. I get this error when I call a nonexistent function:

Function call failed Cannot read property 'args' of undefined

When you call npm start you should be able to pass in --verbose.

There should be a single setting (there are many right now) that gets set to true in order to log a lot more data. We should read this parameter for both spark-protocol and spark-server. I know there are some checks in SparkCore.js that have if (shouldLog) that could use this.

If I repeat a get request to /v1/devices/{deviceid}/{varname} every 1 second, the server stops returning the variable's value and starts returning {"error":"\"value\" argument is out of bounds","ok":false} instead in under 5 minutes.

This error only impacts the device that was being polled, not any other devices on the server. Leaving the device idle for several minutes does not fix the issue. In current testing, the issue remains until the spark-server process is restarted.

I have not noticed any related events in the server console yet.

Please let me know if I can provide any additional information or testing to help diagnose this.

Have fun with this one @AntonPuko

To ease automation for some use-cases, it would be useful to be able to take ownership of a device as an "administrator" without requiring the original owner to release it first.

When doing local compile you can now select which version to compile against

::ffff:192.168.0.250 - - [29/Mar/2017:02:33:02 +0000] "GET /v1/build_targets?featured=true HTTP/1.1" 404 9 "-" "node-superagent/2.3.0"

I'm not sure what the actual data looks like but you should be able to call the public cloud to see it.

add support for webhook rejectUnauthorized option. something about SSL certificate validation of the target URL

I've gotten myself a simple Docker image up and am trying to run this clone of spark-server inside it.

It's running but I get the errors pasted below. (note 2 devices here I think one is a core, one is a photon - I'll isolate later):

I have been running these devices quite happily using this single tweak to the old spark-server : https://github.com/lbt/spark-server/commits/lbt-working-master

So from an identical copy of that working setup (with the same core_keys/ and users/ files) I essentially checked out and ran the new code.
It starts with :

$ docker run -p 8080:8080 -p 5683:5683 spark-server 
Your device server IP address is: 172.17.0.2
Server started on port: 5683
express server started on port 8080

but then:

Connection from: ::ffff:10.0.0.102 - Connection ID: 189
Handshake failed:  error:0407109F:rsa routines:RSA_padding_check_PKCS1_type_2:pkcs decoding error { cache_key: '_189', deviceID: null, ip: '::ffff:10.0.0.102' }
1 : Device disconnected: Error: error:0407109F:rsa routines:RSA_padding_check_PKCS1_type_2:pkcs decoding error { cache_key: '_189', deviceID: '', duration: undefined }
Device startup failed: error:0407109F:rsa routines:RSA_padding_check_PKCS1_type_2:pkcs decoding error
Connection from: ::ffff:10.0.0.125 - Connection ID: 190
Handshake failed:  error:04065084:rsa routines:RSA_EAY_PRIVATE_DECRYPT:data too large for modulus { cache_key: '_190', deviceID: null, ip: '::ffff:10.0.0.125' }
1 : Device disconnected: Error: error:04065084:rsa routines:RSA_EAY_PRIVATE_DECRYPT:data too large for modulus { cache_key: '_190', deviceID: '', duration: undefined }
Device startup failed: error:04065084:rsa routines:RSA_EAY_PRIVATE_DECRYPT:data too large for modulus

I'm not an expert at nodejs so please let me know what you need to know.
Below is my annotated Dockerfile which defines the container - it should give you all the server info you need :)

I'm pretty sure the devices are on stock v0.6.0 firmware from a local build.

# Use the official nodejs v6.10.0 LTS container as a starting point
FROM node:boron

# Create an 'app' directory somewhere in the container
RUN mkdir -p /usr/src/app
WORKDIR /usr/src/app

# Install app dependencies by copying in the package and using the preinstalled
# npm
COPY package.json /usr/src/app/
RUN npm install

# Bundle app source from outside the container to inside
COPY . /usr/src/app
# Do the babel thing to handle new features
RUN ./node_modules/.bin/babel src/ -d lib

# The access ports will be:
EXPOSE 8080 5683

# Now run the babel'ised main.js (is this the right command?)
CMD [ "node", "lib/main.js" ]

The _id is an implementation detail of MongoDB. Since We are overriding it for devices, I am running into issues with my test script.

I think it's fine to use _id for webhooks/users but for devices the _id should be handled by Mongo. The device ID comes from the device itself and isn't a trustworthy source for generating an ID. In my case, I'm getting errors when I send the public key + device ID.

Is the dev branch currently broken? I just wanted to give the server a quick try after reading about it on community.particle.io, but I'm getting the following error message (see below). Am I doing something wrong?

npm start

> [email protected] start /Users/Test/particle.io/spark-server
> nodemon --exec babel-node ./src/main.js --watch src --watch ../spark-protocol/dist --ignore data

[nodemon] 1.11.0
[nodemon] to restart at any time, enter `rs`
[nodemon] watching: /Users/Test/particle.io/spark-server/src/**/* ../spark-protocol/dist
[nodemon] starting `babel-node ./src/main.js`
module.js:472
    throw err;
    ^

Error: Cannot find module '../../third-party/specifications'
    at Function.Module._resolveFilename (module.js:470:15)
    at Function.Module._load (module.js:418:25)
    at Module.require (module.js:498:17)
    at require (internal/module.js:20:19)
    at Object.<anonymous> (/Users/Test/particle.io/spark-server/node_modules/spark-protocol/dist/lib/FirmwareManager.js:67:23)
    at Module._compile (module.js:571:32)
    at Module._extensions..js (module.js:580:10)
    at Object.require.extensions.(anonymous function) [as .js] (/Users/Test/particle.io/spark-server/node_modules/babel-register/lib/node.js:152:7)
    at Module.load (module.js:488:32)
    at tryModuleLoad (module.js:447:12)
    at Function.Module._load (module.js:439:3)
    at Module.require (module.js:498:17)
    at require (internal/module.js:20:19)
    at Object.<anonymous> (/Users/Test/particle.io/spark-server/node_modules/spark-protocol/dist/server/DeviceServer.js:63:24)
    at Module._compile (module.js:571:32)
    at Module._extensions..js (module.js:580:10)
    at Object.require.extensions.(anonymous function) [as .js] (/Users/Test/particle.io/spark-server/node_modules/babel-register/lib/node.js:152:7)
    at Module.load (module.js:488:32)
    at tryModuleLoad (module.js:447:12)
    at Function.Module._load (module.js:439:3)
    at Module.require (module.js:498:17)
    at require (internal/module.js:20:19)
    at Object.<anonymous> (/Users/Test/particle.io/spark-server/node_modules/spark-protocol/dist/index.js:24:21)
    at Module._compile (module.js:571:32)
    at Module._extensions..js (module.js:580:10)
    at Object.require.extensions.(anonymous function) [as .js] (/Users/Test/particle.io/spark-server/node_modules/babel-register/lib/node.js:152:7)
    at Module.load (module.js:488:32)
    at tryModuleLoad (module.js:447:12)
    at Function.Module._load (module.js:439:3)
    at Module.require (module.js:498:17)
    at require (internal/module.js:20:19)
    at Object.<anonymous> (/Users/Test/particle.io/spark-server/src/defaultBindings.js:5:1)
    at Module._compile (module.js:571:32)
    at loader (/Users/Test/particle.io/spark-server/node_modules/babel-register/lib/node.js:144:5)
    at Object.require.extensions.(anonymous function) [as .js] (/Users/Test/particle.io/spark-server/node_modules/babel-register/lib/node.js:154:7)
[nodemon] app crashed - waiting for file changes before starting...

I can create multiple users with the same email right now. We should validate that there isn't an existing user before creating a new one.

I also see that I can claim the same device multiple times.

I've set up a load balancer with two spark-servers attached. Commands to the devices are being routed through the load balancer. I'm running into a problem where the device itself is connected to server1, but external requests are being split between server1 and server2. If the request goes into server2, it thinks the device is not connected.

I was thinking that a solution would be have server2 point to server1's data directory - is there any way to do this?

EDIT:
Thinking about this some more....don't think simply sharing the data folder would work, right? I was playing around with the clustering options here https://github.com/Unitech/pm2 and was able to get multiple workers going. But, the same problem....if I do a GET /v1/devices, the device will only report back as connected if the GET request was sent to the same fork that the device is really 'connected' to. Is the app not stateless?

Would you say this would work well for a production product?
What would be a reasonable quantity of interactions from devices that this could handle on a modern computer.

Having a dependency on OpenSSL and Python sucks. A lot of people run into trouble when trying to use these libs.

Use this instead:
https://github.com/rzcoder/node-rsa

Hi,

I have a project in my company, and we need to use a local cloud for Particle Photons.
Because need all time the availability of exchange messagens between some Photons.

All how to in the internet for install in Raspberry Pi (2) not work for Spark Server.

I need know a simple but functionall how to for install in Raspberry Pi 2 or in any plataform (recommended) .
And I need know if need just the Spark Server installed, or Spark Protocol (Particle).
Some erros I have encontered is about node deprecated functions. All not explain the exact version for this actual Spark Server.

Thanks
Adio

I've thought this over again and I really do not like that the data we are creating is being saved inside the src folder. I also think it will cause confusion as more people start using this repo.

I don't think that we've covered them all. We need to finish some in DeviceController as well

I also wonder if we should be using a real mocking framework instead of hacking some mocked objects..?
We could write better tests with the mocked objects like checking if functions had been called or getting rid of the serialized functions -- we can make repositories return fake data.

We need optional fields in the settings for SSL.
http://stackoverflow.com/questions/11804202/how-do-i-setup-a-ssl-certs-for-an-express-js-server

On the particle cloud, I am able to pass variables to webhooks (requestType and headers). When pointing to my local server, when I try to add below webhook, I get "Error wrong requestType". This is simple enough to change and it should be no problem for me to just hard code the request type. However I need to be able to send a variable Authorization token.

Is this something that's changing with your project or am I just missing something? The webhook I'm trying to set up worked on the particle cloud before.

spark-server responds with:

webhookError: Error: Unauthorized

webhook:

{
  "event": "myevent",
  "url": "http://myapi.com/v1/{{path}}",
  "requestType": "{{request}}",
  "headers":
    {
        "Content-Type": "application/x-www-form-urlencoded",
        "Authorization": "{{api_key}}"
    },
  "responseTopic": "hook-response/api_{{SPARK_CORE_ID}}",
  "responseTemplate": "{{error}}:{{message}}",
  "form":
    {
        "var1": "{{var1}}",
        "value": "{{value}}"
    },
  "mydevices": true
}

Implement webhooks to match the Particle spec.

Make sure that if I add/remove a webhook from the CLI that it removes the subscription from the event listeners.

event:

event: sensors
data: {"coreid":"20003b000d47343432313031","data":"{\"t1\":0.000000, \"t2\":0.000000, \"t3\":0.000000, \"t4\":0.000000, \"flow1\":0.000000, \"lux\":19.095816  }","published_at":"2017-01-14T11:10:09.470Z","ttl":60}

webhook:

{
  "event": "sensors",
  "url": "https://api.pushover.net/1/messages.json",
  "requestType": "POST",
  "form": {
    "user": "XXX",
    "token": "YYY",
    "title": "TEST",
    "message": "{{ SPARK_CORE_ID }} lux: {{ lux }}"
  },
  "ownerID": "b20fd7ddd647fcb84d0f859c803e01e2",
  "created_at": "2017-01-14T08:41:01.106Z",
  "id": "df1bd069-d87b-4209-8294-8e9246d74a0a"
}

it happens also with plain data not only with json format

@AntonPuko @wdimmit @durielz

I'm thinking of using a IoC framework in order to easily swap out the different repositories used here. This change will cascade down into the protocol layer.

I think the change will make swapping out certain implementations much easier but I want some feedback before I start as this is a super opinionated change.

An example of why we should do this is:
Right now we have concrete implementations of saving device attributes and server keys to the local file storage. Yes you can change the path via settings but what if you want to fetch these via HTTP or using some storage API (Azure) or whatever the AWS storage is.

By moving over to a framework like this we would be able to swap out implementations by adding a single line and the configuration would cascade down.

Here is one I'm looking at because it's fairly simple and light-weight.

If my device disconnects from network i get:
Caught exception: TypeError: Cannot read property 'message' of undefined TypeError: Cannot read property 'message' of undefined at Socket.<anonymous> (/node_modules/spark-protocol/lib/clients/Device.js:175:66)

changing in client/Device.js in protocol the line 134 from
error: Error): void => this.disconnect(socket timeout ${error.message}),
to
error: Error): void => this.disconnect(socket timeout ${error}),
then i have the correct disconnect message

Sent particle keys doctor [devid] and keys appear to have sent successfully. The photon restarts but blinks a very fast cyan with a red blink every so often. On the local server:

::ffff:10.10.10.107 - - [29/Jan/2017:21:31:34 +0000] "POST /v1/provisioning/[devid] HTTP/1.1" 200 128 "-" "-"

Handshake failed: error:0407109F:rsa routines:RSA_padding_check_PKCS1_type_2:pkcs decoding error { cache_key: '_64', deviceID: null, ip: '::ffff:10.10.10.117' }
1 : Core disconnected: Error: error:0407109F:rsa routines:RSA_padding_check_PKCS1_type_2:pkcs decoding error { cache_key: '_64', deviceID: '', duration: undefined }
Connection from: ::ffff:10.10.10.117 - Connection ID: 64

Connection ID increments about once every 2 seconds.

I can revert keys back to the default particle public keys, and the photon seems to work as normal (breathing cyan and connects to particle server just fine). Not sure if this is related, but I also noticed that if I do a particle identify when config'd to the particle server, I get the normal device ID and firmware response, but if I do the same particle identify when config'd to the local server, I get device ID and

Serial err: Error: Opening COM3: Access denied
Serial problems, please reconnect the device.
Unable to determine system firmware version