Error with handshake process #295
Comments
|
What particle device are you using? |
|
It's an electron, and yes! I did the step to change the profile ( I had to change the command slightly to work tough: particle keys server default_key.pub.pem --host IP_ADDRESS --port 5683 --protocol tcp and also particle keys protocol --protocol tcp |
|
By the way, I'm using the latest particle CLI version (2.3.0) |
|
There are two things I can think of
|
|
I did the whole process few times, but I can check again! |
|
Ok, I think I just need to use the original version of this - usrz/ec-key@23f4dd3 I'm guessing the dependencies are just out of date now. |
|
I've changed Is there anything I'm missing? |
|
I updated the dependency. You'll need to pull the latest version of this repo and |
|
I've seen that spark-protocol repo was updated, but not this. Shouldn't this package.json be updated with the new ec-key ^0.0.4 too? I can't update properly. It keeps downloading the spark-protocol without your last changes |
|
Doh -- I didn't push. It's updated now. |
|
Tried the with the updated version but still fails with the same error. Also I've moved to node This is the "updated" error log:
|
|
The only thing I can think of is that the key is improperly formatted. Can you run a test script with |
|
There the test with the two keys. The one in deviceKeys.db (also matches the generated XXXXX_rsa_new.pub.pem) and the one received from the device during handshake. const ECKey = require('ec-key');
device_db_pem = "-----BEGIN PUBLIC KEY-----\nMFkwEwYHKoZIzj0CAQYIKoZIzj0DAQcDQgAEaUXZhpYnCARKEx1FGFrOgu8Dgyb5\n//Drwxg2oR8LkP37MDj7ESmj78PdBqD6PeNmvBMQg5Z7NQ8saRDxX1h50g==\n-----END PUBLIC KEY-----\n"
device_sent_pem = "-----BEGIN PUBLIC KEY-----\nMIGfMA0GCSqGSIb3DQEBAQUAA4GNADCBiQKBgQBThDnRSoUGN20VitFNf4Kj1wWh\nIdiJvFJ4CPj2oAoGCCqGSM49AwEHoUQDQgAEaUXZhpYnCARKEx1FGFrOgu8Dgyb5\n//Drwxg2oR8LkP37MDj7ESmj78PdBqD6PeNmvBMQg5Z7NQ8saRDxX1h50v//////\n/////////////////wIDAQAB\n-----END PUBLIC KEY-----\n"
var key = new ECKey(device_db_pem, 'pem');
console.log(key)
var key2 = new ECKey(device_sent_pem, 'pem');
console.log(key2)
/* The second throws the error:
D:\spark-server>node test.js
D:\spark-server\node_modules\asn1.js\lib\asn1\base\reporter.js:84
throw err;
^
ReporterError: Failed to match tag: "objid" at: ["algorithmIdentifier"]["parameters"]
at DecoderBuffer.error (D:\spark-server\node_modules\asn1.js\lib\asn1\base\reporter.js:78:11)
at DERNode.decodeTag [as _decodeTag] (D:\spark-server\node_modules\asn1.js\lib\asn1\decoders\der.js:71:19)
at DERNode.decode [as _decode] (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:341:25)
at decodeChildren (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:378:15)
at Array.forEach (<anonymous>)
at DERNode.decode [as _decode] (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:375:22)
at decodeChildren (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:378:15)
at Array.forEach (<anonymous>)
at DERNode.decode [as _decode] (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:375:22)
at DERNode.decode [as _decode] (D:\spark-server\node_modules\asn1.js\lib\asn1\base\node.js:280:47) {
path: '["algorithmIdentifier"]["parameters"]'
}
*/ |
|
Ok -- that's super helpful. It's definitely a different format. I'm wondering if it's just an RSA key instead of an EC key. This makes me think that I don't need the ec-key parsing code since I'm not using UDP. |
|
I'm not completely sure of how is this issue related because I always use --protocol tcp in all particlecli commands. Also, that issue is closed, so not sure how they finally implemented it. |
|
What I'm saying is that I implemented Electron support before they fixed that bug. In the past the Electron always sent ecc keys for UDP and TCP. Now that it's fixed, I can just use the RSA key code. I'll try to find my electron and see if I can get this fixed tonight. |
|
Let me know if this works -- I couldn't find my electron so I am just guessing at a fix. |
|
It now seems to compare the keys properly but still the keys doesn't match: 07:36:51.287Z INFO DeviceServer.js: New Connection
07:36:52.332Z ERROR Handshake.js: Handshake failed (cache_key=_31, deviceID=XXXXXXXXXXXXXXXXXXXXXXXX, ip=::ffff:XX.XX.XX.XX)
Error: key passed to device during handshake doesn'tmatch saved public key: XXXXXXXXXXXXXXXXXXXXXXXX
at Handshake._callee5$ (D:\spark-server\node_modules\spark-protocol\dist\lib\Handshake.js:388:21)
at tryCatch (D:\spark-server\node_modules\regenerator-runtime\runtime.js:62:40)
at Generator.invoke [as _invoke] (D:\spark-server\node_modules\regenerator-runtime\runtime.js:296:22)
at Generator.next (D:\spark-server\node_modules\regenerator-runtime\runtime.js:114:21)
at step (D:\spark-server\node_modules\babel-runtime\helpers\asyncToGenerator.js:17:30)
at D:\spark-server\node_modules\babel-runtime\helpers\asyncToGenerator.js:28:13
07:36:52.333Z INFO Device.js: Device disconnected (cache_key=_31, deviceID="", disconnectCounter=1)
07:36:52.334Z ERROR DeviceServer.js: Device startup failed (deviceID=null)
Error: key passed to device during handshake doesn'tmatch saved public key: XXXXXXXXXXXXXXXXXXXXXXXX
at Handshake._callee5$ (D:\spark-server\node_modules\spark-protocol\dist\lib\Handshake.js:388:21)
at tryCatch (D:\spark-server\node_modules\regenerator-runtime\runtime.js:62:40)
at Generator.invoke [as _invoke] (D:\spark-server\node_modules\regenerator-runtime\runtime.js:296:22)
at Generator.next (D:\spark-server\node_modules\regenerator-runtime\runtime.js:114:21)
at step (D:\spark-server\node_modules\babel-runtime\helpers\asyncToGenerator.js:17:30)
at D:\spark-server\node_modules\babel-runtime\helpers\asyncToGenerator.js:28:13I've did the complete process with the updated code to re-generate the device/database keys and be sure that it's all set properly. |
|
Do you have something other than an electron to test with? I just want to make sure that it's a server issue and not with the setup process. |
|
Yes! I've tried with Photon and it works! But still not working with Electron. I also checked a Boron but it doesn't even connect, so might be an error of UDP/TCP. |
|
So the |
|
All devices have ecc key in the database (except photon). I'm pretty sure I've deleted it before the last update. |
|
I've did it again cleaning the database but it still shows Database: Keys generated in the computer: https://we.tl/t-C9NqPlanxq Commands used:
|
|
It looks like you're doing everything correctly. I'll need to find my electron to debug this. |
|
I got my Electron set up this morning but did not hook it back up to a network so I just tested and verified that it would send an RSA key instead of an ECC key. Try this: I'm going to open an issue with Particle CLI but I doubt they will fix it on their end. |
|
I executed your commands and it connected properly! RSA in the database and everything seems OK about the keys. But there's a problem during the protocol initialization:
I've done the process from the beginning but always happen the same. I tried DeviceOS 0.7.0 and 1.2.1. |
|
Well, that's progress! For some reason it seems like the |
|
Yes! Thank you for all your support! BTW: Which version do you use for the ParticleCLI / DeviceOS? It seems weird that all this errors are happening only to me, :( |
|
I upgraded to the latest. The reason it's happening to you is that you're using an Electron. Nobody really uses that and hosts their own cloud. |
|
So I connected my Electron to my server and don't see the |
|
Maybe it's related to 1.2.1 version? Not sure about the |
|
Could be. My Electron is the latest. Can you try flashing your Electron with |
* Added npm functions for example build and run, changed eventProvider sample * Changed sample * changed ignore on npm build:examples * patched sample to working code * add prettier, add gitAttributes to hide dist in commit diffs * Changed Infos in package.json * Added Bunyan Logger * added last NIT from last PR * Added Changes from jlkalberer * Added Changes AntonPuko * Added WebHookLogger Log Line * remove gitAdd from npm scripts, its lint-staged own command * update package-lock.json, closes #229 * remove wrong logger.warn in webhookManager * remove wrong logger.warn in webhookManager * update package-lock.json, closes #234 * update migrateScript, so it saves dates correctly. * passing not json event.data as sting for webhooks as form-ulr-encoded * update build * stringify device variables * try/catch around device attributes as JSON deserializer was breaking on existing devices. * Update to newer version of spark protocol * build deviceAttributesDatabaseRepository.js * Starting work on products API. Still need to add in device endpoint and update the firmware manager so it will auto-update devices. * fix adminUser creation * Finished API for managing firmware/devices/products. Now we need to actually add code for flashing the correct firmware when the device connects. * Added changes so that when a product firmware is updated, it sends an event to the device server to start updating all connected devices. * Rebuilt dist * Fixed DB so it fetches correctly for mongo. * fix variables loss on update device name. * remove console.logs; add webhookLoggerStub * add bunyan to start scripts * add DeviceManager to ProductsController bindings * provide webhookData as querystring for GET requests * update package-lock.json to get last spark-protocol changes. * remove comments * add deviceManager assignment in productsController, closes #245 * update package-lock.json for spark-protocol change: Brewskey/spark-protocol@3472b1a closes #244 * update package-lock.json for spark-protocol change: Brewskey/spark-protocol@b5ede6d * update package-lock.json for spark-protocol change: Brewskey/spark-protocol@2e1dde3 * update package-lock.json for spark-protocol change: Brewskey/spark-protocol@6e83a46 * product api fixes * revert back casting product_id to string on production creation, use helper function for getting numeric productID from IDOrSlug * add equals to 0 check for right platform_id filtering. * fix _formatProduct in ProductDatabaseRepository * add parseInt for firmware version comes from route params. * formatDeviceAttributes, add product_id to returned product devices props * build files * use isNaN in getByIDOrSlug * fix flashProductDevices method, fix csv parsing * save product_id instead product.id in productDevice * cast productFirmware buffer when save to/ get from db. * update flow-bin, fix some product related flow errors, there are still some * fix bugs, add temporary hack for getting right buffer from different dbs * add devicePing endpoint * disable userFilter for admin in eventsController * return keep alive for event stream * addproductDevices small fixes, rebuild files * Fixed bindings...? * Updated products controller to flash when devices are added to an existing product. Fixed some flow errors. * Always flash new devices when they are added to a product. * Not sure why these changes weren't checked in * bump package-lock.json * bump package-lock.json again with rebuilt spark-protocol * bump spark-protocol verison in package-lock for that change Brewskey/spark-protocol@566fb0a * Updated README FirmwareCompilationManager doesn't depend on external files. Flow fixes. * update package-lock.json for that change Brewskey/spark-protocol@3f2bf42, fix #265 * bump spark-protocol version * set current === false for existing released firmware on update/addnew firmware * remove missing api features from readme * fix current is undefined * Fixed updates on product firmware Fixes #266 * Rebuild * bump package-lock.json * fix some products endpoint payloads under v2 routes * remove page and pageSize from db.find() * fix default skip/take value in mongodb * Updated ec-key Now using yarn * Update yarn.lock * and again.. * ............ * This should be the one * Update yarn lock * parse skip and take to integer for mongo * case insensitive search for getByIDs, should close #279 * add paging for v2 getFirmares, add countFirmwares, fix find() with take = 0 return all entities * remove test endpoint * remove default take value, better skip/take checks * move firmwares endpoints to its own controllers * revert case sensitive search for getManyFromIDs for now * Fixed issue with adding products. If the device had never connected to the cloud, the platformID wasn't set on DeviceAttributes. * Update spark-protocol version * Update yarn.lock * ... * add getFirmware route * fix event stream format under eventsControllerV2 * Fixes to productdevices query Always `toLowerCase` when working with particle device IDs. There are cases where we query/save without doing this so I'll write a migration script later. * Fix ProductDevices return value so it always returns `id`. If attributes are null, it doesn't return the value. * remove deviceAttributes check on delete productDevice * Update firmware queries so device_count queries the database instead of storing the value. * Fix bindings. The ProductFirmwareController didn't work because it wasn't getting all the parameters injected. * The DeviceAttributeDatabaseRepository was patching instead of updating. We need to get all the existing data in order to know the device product firmware version. * Trying to get the firmware version into productdevices... * doh.. forgot to use `await` * Return `notes` and other fields from ProductDevices API * You should be able to toggle quarantined devices even if they haven't connected yet. * send /n on sse connection start * Updated yarn dependencies * Updated spark-protocol version * Updated spark-server dependency. * Updated dependencies * Fixing logging in WebhookManager * Updating logging * Updating logging to be more consistent and include deviceID if possible * Fix WebhookManager.test.js * Fixing exception in website manager * More logging fixes * More webhook logging for errors. * Add additional SSE header * Update protocol version * Update spark-protocol that supports "@" in binary filenames. * Update events controller to return the proper format :/ * ... * Update SSE to return UTF-8 encoding header. * Removing changes to event stream. We want the default message type to be used so we only have to subscribe to one thing. * revert keepalive. * There was an unimplemented function on this repository.. this will fix firmware updates. * Fixes #290 * Add support for name or device ID on endpoints for #293 * One more stab at #293 * Fixing flow types * Upgraded spark dependency * Update README.md * Update README.md * Upgraded yarn.lock * Upgrade spark-protocol **Make sure to run `yarn upgrade-firmware` in order to get OTA update fixes** * Upgraded dependencies to get new fix for flashing larger binaries * Updated ec-key and rebuilt * Hopefully fixes #295 * Update README.md * Update README.md * Upgrade spark-protocol * Fixed all flow types Server seems to be running but I'll need to test * Fixing scripts for npm * Updated to latest spark-protocol * Updated dependencies * Update reference to spark-protocol. Fixes crypto errors. * Updated CORS to run for all requests * Adding index generation * Updated server to use fixed spark-protocol Added support for a settings.json file to override settings. * Rebuild server Updated link in readme * Fixing #298 * Fixed events controller crashing the whole server and added some logging. * build server * Upgrade dependencies * Remove extra logging * Upgrade spark-protocol for connection throttling. * Upgrade spark protocol for OTA update fixes. * Update spark-protocol * Update spark-protocol * Upgrade spark-protocol * update spark-protocol * upgrade spark-protocol * Fix coap functions * Update spark-protocol * Update spark protocol * Update spark-protocol * update spark-protocol * Fixed binding for config Updated spark-protocol * Update packages * Remove postinstall step * Update dependencies * Update mongodb version * Fix mongo collection call * Get db from client * Fix database * Updated spark-protocol * update deps * update * bump * bump * bump * bump * bump * bump * bump * bump * bump * bump * bump * bump * bump * Migrated everything to typescript * Add improved logging * Redacted token from authorization header. * Improve types * bump * bump * Fix mkdirp * Delete generated files * Migrate to lerna + workspaces * Working on publish workflow * Trying to fix npm ci in GH action * fix commands * Remove remote e2e tests * Make private false * public access * only publish dist folder * Publish spark-protocol too * Bump version * fix publish.yml * fixing exports * Remove crc check when updating firmware * Bump packages * Move particle-collider * Fix webhooks Bump modules for publish --------- Co-authored-by: Andreas <[email protected]> Co-authored-by: Andreas Häferer <[email protected]> Co-authored-by: Anton puko <[email protected]> Co-authored-by: Anton Puko <[email protected]> Co-authored-by: AntonPuko <[email protected]>
Hi,
I've just tried the steps to build but I'm having troubles during the authentication process.
The error comes when the particle device tries to connect to the cloud but if I look to deviceKeys.deb, the data for that device is properly stored:
The error is when comparing
publicKey.equals(deviceProvidedPem)at Handshake.js:383 from spark-protocol (I posted the issue here because I don't really know what is the root cause)The error log shows this:
I'm using node
v8.11.1, npm5.6.0and yarn1.22.4, and tested the server on Ubuntu&Windows.Thank you
The text was updated successfully, but these errors were encountered: