ndidplatform / api Goto Github PK

/ndid/initNDID, if success, will respond with [true, <height>]. Instead, it should respond with success status (true/false) only.

As I have investigated, the issue seems to be caused by

Since tendermint.transact() returns an array, the line should destructure the returned array and return only success status instead of the whole array.

So it should be something like:

const [ result, height ] = tendermint.transact(
...
init = result;
return init; 

'bignum' dependency (C++ binding) is unable to compile successfully when using with NodeJS 10

I have two IDP, AS nodes and RP create request with data request (min_idp=1, min_as=2)
two IDP nodes are respond request at the same time and final callback request status to RP has duplicate response_valid_list like this.

response_valid_list: [ { idp_id: 'idp2', valid_proof: true, valid_ial: true },
{ idp_id: 'idp2', valid_proof: true, valid_ial: true } ]

debug: Query MQ destination for RP { rp_id: null }

Environment

• NDID, smart-contract v4.0.0
• NDID, api v3.0.1
• Red Hat Enterprise Linux Server 7.7

[docker_mq-rp_1 exited with code 139
[docker_mq-as_1 exited with code 139
[mdocker_mq-idp-1_1 exited with code 139
[api-ndid_1] Segmentation fault (core dumped)
[docker_mq-idp-2_1] exited with code 139

docker compose up logs:
api-compose-up-with-mq-error.log

Reproduce Steps

• Start up a new node in docker
• Query to function GetNodeMasterPublicKey via Tendermint RPC

Current Result

• Get {"master_public_key":""}

Expected Result

• Get {"master_public_key":"<key_in_pem_format>"}

Cause

• Typo in start up script start-api.sh when setting path to MASTER_PUBLIC_KEY_PATH

from mq.js

export const defaultMqBindingPort = (() => {
  if (process.env.ROLE === 'as') return 5557;
  if (process.env.ROLE === 'idp') return 5555;
  if (process.env.ROLE === 'rp') return 5556;
})();

If we not set MQ_BINDING_PORT before start process as ndid role, the process will be crashed with

/home/circleci/project/node_modules/zeromq/lib/index.js:451
  this._zmq.bindSync(addr);
            ^
Error: Invalid argument
    at exports.Socket.Socket.bindSync (/home/circleci/project/node_modules/zeromq/lib/index.js:451:13)
    at Object.<anonymous> (/home/circleci/project/src/mq/index.js:12:17)
    at Module._compile (module.js:652:30)
    at Object.Module._extensions..js (module.js:663:10)
    at Module.load (module.js:565:32)
    at tryModuleLoad (module.js:505:12)
    at Function.Module._load (module.js:497:3)
    at Module.require (module.js:596:17)
    at require (internal/module.js:11:18)
    at Object.<anonymous> (/home/circleci/project/src/core/rp.js:10:1)
    at Module._compile (module.js:652:30)
    at Object.Module._extensions..js (module.js:663:10)
    at Module.load (module.js:565:32)
    at tryModuleLoad (module.js:505:12)
    at Function.Module._load (module.js:497:3)
    at Module.require (module.js:596:17)
    at require (internal/module.js:11:18)
    at Object.<anonymous> (/home/circleci/project/src/routes/rp.js:4:1)
    at Module._compile (module.js:652:30)
    at Object.Module._extensions..js (module.js:663:10)
    at Module.load (module.js:565:32)
    at tryModuleLoad (module.js:505:12)

Issue

When starting a new tendermint node, it will sync/replay all the transactions.

Possible Solution

Track chain sync status by checking the result returned from calling /status to tendermint.

Add CALLBACK_RETRY_TIMEOUT as an environment variable for configurable callback retry timeout.

This config applies ONLY to callback retries that have not set a pre-retry hook function (shouldRetry function). The timeout value is in seconds. Default is 600 seconds or 10 minutes.

When AS handle request callback with status 200 (with data), they also provide the callback_url and reference_id for callback, to know if blockchain method signData is success, but no callback is fired.

RP should process request status update only once if there are multiple transactions with the same request ID in the same block. Related to issue #27.

Should handle error when sending an IdP response to a closed or timed out request and send back a correct error code/message.

Currently, messages sending over message queue are in JSON string format which causes the payload to be printable string. Hence, the payload is encoded in base64 string which makes the size bigger than it should.

It is more suitable to send data in raw bytes rather than bytes of UTF-8 string by using protobuf.

Scenario:
Create Request with
min_idp: 1
idp_id_list: [idp1, idp2]
mode: 3

When reject at idp1 and accept at idp2, the request was rejected.

I think this case the request status should be completed by 1 IdP accept.

When request for authentication status is completed, Request is "closed=false".
It should be "closed=true".

Issue

Having to wait for ABCI app output:

Commit
Commit

before starting api server.

Possible Solution

Listen for 'NewBlock' event from tendermint and check if height >= 2 (Genesis block committed and signed) before receiving/processing any requests.

When creating a new identity, the final request status is always "pending". It should be "completed".

Currently, request timeout trigger uses setTimeout() which support timeout up to only 2147483647 milliseconds (or about 24 days).

Need to come up with another way to trigger a timeout to support long scheduled request timeout.

Throw an error if an expected value or object property returned from external crypto service is null or undefined.

Create API to toggle the crypto service choice on the fly.

I mentioned this to @taneekpek at Barcamp. I would like to propose that bcrypt is used as a hash algorithm instead of sha256.

To prevent downtime by accident (for hot standby).

Expose from smart contract

When GET /as/service/:service_id, response format is
{
"service_id": "string",
"url": "string"
"service_name": "string"
}

It should be (Swagger hub)
{
"min_ial": 1.1,
"min_aal": 1,
"url": "string"
}

If service is registered by NDID but not offered by any AS it correctly return 200 with empty array.
But if it's not registered by NDID it should return 404 but now return empty array.