Reputation: 218
I tried two different tutorials:
This one to deploy docker images locally: https://github.com/hyperledger/fabric-samples I tried the fabcar sample. Everything works fine. I'm able to enroll the admin, register a user and invoke transactions using this user identity. Nice.
The second one to deploy a fabric composer playground on bluemix: https://ibm-blockchain.github.io/setup/ I was able to get into the composer playground and deploy business network examples. I'm also able to upload and deploy my own business networks I designed locally with composer-cli. I exposed the network over a Rest Server with admin rights in order to test the transactions. All works fine.
Now it comes to my problem. I want to combine both examples (just for fun, learning by doing:)). I want to deploy the fabcar samples network to the kubernetes environment on bluemix and interact with the network using the fabric node sdk like in example 1. So I made following changes in the scripts/yaml files of example 2:
in the create_all.sh. line 41 (for peer 1) and line 45 (for peer 2). I changed the values of the env variables CHAINCODE_NAME and CHAINCODE_VERSION
echo "=> CREATE_ALL: Running Install Chaincode on Org1 Peer1" CHAINCODE_NAME="fabcar" CHAINCODE_VERSION="1.0" MSP_CONFIGPATH="/shared/crypto-config/peerOrganizations/org1.example.com/users/[email protected]/msp" PEER_MSPID="Org1MSP" PEER_ADDRESS="blockchain-org1peer1:30110" create/chaincode_install.sh
in the chaincode_install.yaml.base line 16, I changed the path of the chaincode. I forked the https://github.com/hyperledger/fabric and added the fabcar.go chaincode from example 1 to the new repository https://github.com/bigbelette/fabric
command: ["sh", "-c", "git clone https://github.com/bigbelette/fabric $GOPATH/src/github.com/bigbelette/fabric/ && peer chaincode install -n ${CHAINCODE_NAME} -v ${CHAINCODE_VERSION} -p github.com/bigbelette/fabric/examples/chaincode/go/fabcar/"]
in the create_all.sh. line 49. I changed the values of the env variables CHAINCODE_NAME and CHAINCODE_VERSION:
CHANNEL_NAME="channel1" CHAINCODE_NAME="fabcar" CHAINCODE_VERSION="1.0" MSP_CONFIGPATH="/shared/crypto-config/peerOrganizations/org1.example.com/users/[email protected]/msp" PEER_MSPID="Org1MSP" PEER_ADDRESS="blockchain-org1peer1:30110" create/chaincode_instantiate.sh
In the chaincode_instantiate.yaml.base, I made the parameter list empty:
command: ["sh", "-c", "peer chaincode instantiate -o blockchain-orderer:31010 -C ${CHANNEL_NAME} -n ${CHAINCODE_NAME} -v ${CHAINCODE_VERSION} -c '{\"Args\":[\"\"]}'"]
The deployment on the cluster works without an error. In the fabcar examples node js file enrollAdmin, regiserUser, query and invoke, I changed the channel name (from mychannel to channel1) and the IPs and ports to match to the online kubernetes services:
enrollAdmin.js and registerUser.js work fine so I can get the certificates and store it in the key store. But when I try to invoke a transaction, i get the following error:
It seems the CA which gave me the certificates is not recognized by the peers. But there is only one CA in my network!? Or is it a problem in the MSP definition (something I still dont understand well).
Thanks for your help.
Upvotes: 0
Views: 725
Reputation: 218
I finally found the error.
My network has 1 CA (the root CA, named ca) and 2 "sub" CAs (ca1 und ca2). I was enrolling admin and user directly by the root CA because I didnt mention the name of the ca in the parameters of the Fabric_CA_Client class:
fabric_ca_client = new Fabric_CA_Client('http://184.173.5.108:30054', tlsOption, crypto_suite)
Without the name, the used CA will be the default CA (Root CA in my case). But the signature of user certificates by the root CA wont be recognized by the peers. Only the signatures from the last hierarchical CA levels will be accepted. So the solution was to refer the name of the CA:
fabric_ca_client = new Fabric_CA_Client('http://184.173.5.108:30054', tlsOptions , 'CA1', crypto_suite)
It works. Now the user can invoke transactions.
Upvotes: 1