Worker for send whatsapp messages
wa_worker is a service that "listen" through rabbitmq queue (https://www.rabbitmq.com/) and send a message using configured whatsapp credentials (through yowsup implementation https://github.com/tgalal/yowsup), created because need a tool for deliver gerential report every hour using whatsapp, premise is dispatch every request in order, service always available and always must deliver. The product was developed with etcd and docker ecosystem in mind, so it was deployed in coreos.
###Notice:
- The directory wa_worker/scheduler will be separated to a different project with a web client interface because are not related with wa_worker/message_receiver at all
- Using Python 2.7
- Image base for deployment at https://registry.hub.docker.com/u/ivandavid77/centos7dockerfile/dockerfile/
###Steps for using (based on coreos deployment):
###1) Configure rabbitmq
In coreos create a service file like this:
File: rabbitmq@.service
[Unit]
Description=RabbitMq
After=docker.service
Requires=docker.service
[Service]
TimeoutStartSec=0
Restart=always
RestartSec=60
ExecStartPre=-/usr/bin/docker kill cred_mq
ExecStartPre=-/usr/bin/docker rm cred_mq
ExecStartPre=/usr/bin/docker pull rabbitmq
ExecStart=/usr/bin/docker run --name cred_mq -e RABBITMQ_NODENAME=cred-mq -p 8672:5672 rabbitmq
ExecStop=/usr/bin/docker stop cred_mq
[X-Fleet]
Conflicts=rabbitmq@*.service
- Use the official image of rabbitmq at https://registry.hub.docker.com/_/rabbitmq/
Additional we need a discovery unit that announce the location of this service to the cluster:
File: rabbitmq-discovery@.service
[Unit]
Description=Announce RabbitMQ
BindsTo=rabbitmq@%i.service
After=rabbitmq@%i.service
[Service]
Restart=always
RestartSec=60
ExecStart=/usr/bin/bash -c "while true; do echo '{\"host\": \"'$(hostname -i|sed 's/\ $//')'\", \"port\": \"8672\"}' | tr -d '\n' | etcdctl set /services/rabbitmq@%i --ttl 60;sleep 45;done"
ExecStop=/usr/bin/etcdctl rm /services/rabbitmq@%i
[X-Fleet]
MachineOf=rabbitmq@%i.service
This service will create a resouce into etcd that "announce" to cluster our
rabbitmq@1 service, the resource will looks like
/services/rabbitmq@1 {"host":"172.16.200.201", "port":"8672"}
- Its critical that before start the discovery service, the shell command
$ hostname -ireturn the correct ip address, maybe you will need edit your /etc/hosts file:
File: /etc/hosts
172.16.200.202 coreos-node02.crediland.mx
your-ip-addr your-hostname
Check that command gives correct information:
$ hostname
coreos-node02.crediland.mx
$ hostname -i
172.16.200.202
Ok, after that install/start service:
$ fleetctl submit rabbitmq@.service
$ fleetctl submit rabbitmq-discovery@.service
$ fleetctl start rabbitmq@1
$ fleetctl start rabbitmq-discovery@1
Check if loaded
$ fleetctl list-units
UNIT MACHINE ACTIVE SUB
rabbitmq-discovery@1.service 337c3825.../172.16.200.201 active running
rabbitmq@1.service 337c3825.../172.16.200.201 active running
We are ready for next step.
###2) Load wa_worker.
UPDATE: before going further clone the project and edit the file start.sh and
comment out every line inside, only leave untouched the lines:
echo -e ' [*] Loading requirements...\n';
pip install -r $MOUNT_POINT/wa_worker/requirements.txt
and
while true; do
echo ' [*] running...';
sleep 10;
done
Because coreos uses docker, take a look at the image base for this project in: https://registry.hub.docker.com/u/ivandavid77/centos7dockerfile/dockerfile/, the wa_worker service will load github project and mount it into instance created by docker:
File: wa_worker@.service
[Unit]
Description=WhatsAppWorker
After=docker.service
Requires=docker.service
[Service]
TimeoutStartSec=0
Restart=always
RestartSec=60
ExecStartPre=-/usr/bin/docker kill wa_worker
ExecStartPre=-/usr/bin/docker rm wa_worker
ExecStartPre=/usr/bin/docker pull ivandavid77/centos7dockerfile
ExecStartPre=-/bin/rm -fR /home/administrador/tmp/wa_worker
ExecStartPre=/usr/bin/git clone https://github.com/ivandavid77/wa_worker.git /home/administrador/tmp/wa_worker
ExecStart=/bin/sh -c '/usr/bin/docker run --name wa_worker -e SYSLOG_REMOTE=172.16.202.15:514 -e ETCD_ENDPOINT=$(hostname -i) -e ETCD_PORT=4001 -e MQ_INSTANCE=%i -e MQ_SEND_MESSAGE_QUEUE=WA_MESSAGE_QUEUE -e MQ_TASK_MANAGEMENT_QUEUE=WA_MANAGEMENT_QUEUE -e SECRET_KEY=MjAxNTA1MTMxNTE3 -e MOUNT_POINT=/mnt -v /home/administrador/tmp/wa_worker:/mnt/wa_worker ivandavid77/centos7dockerfile /mnt/wa_worker/start.sh'
ExecStop=/usr/bin/docker stop wa_worker
ExecStopPost=/bin/rm -fR /home/administrador/tmp/wa_worker
[X-Fleet]
Conflicts=wa_worker@*.service
Load and install/start the service, verify that its running:
$ fleetctl submit wa_worker@.service
$ fleetctl start wa_worker@1
$ fleetctl list-units
UNIT MACHINE ACTIVE SUB
rabbitmq-discovery@1.service 337c3825.../172.16.200.201 active running
rabbitmq@1.service 337c3825.../172.16.200.201 active running
wa_worker@1.service 8e179c12.../172.16.200.202 active running
Several environment vars are defined inside docker instance in ExecStart line:
SYSLOG_REMOTE :: because i don't want track were are located the instance, better specify where redirect syslog messages
ETCD_ENDPOINT :: wa_worker need know where is their "local" etcd server for ask the location of rabbitmq@ service inside cluster
ETCD_PORT :: same but port number
MQ_INSTANCE :: what instance must be located in "/services/rabbitmq@*" on etcd cluster service? R = instance 1 (rabbitmq@1)
MQ_SEND_MESSAGE_QUEUE :: what queue will be used for listen request for sending messages?
MQ_TASK_MANAGEMENT_QUEUE :: what queue will be used for scheduler service (wa_worker/scheduler will be removed to another github proyect)
SECRET_KEY :: secret that will be used for encrypt/decrypt sensitive config information (whatsapp key/password configuration, email config for fail over, etc.)
MOUNT_POINT :: where was mounted the github project inside docker instance? (check the flag -v /home/administrador/tmp/wa_worker:/mnt/wa_worker ... is in "/mnt", wa_worker need know that!)
###3) Configure wa_worker
Login into wa_worker instance for configuration (according with output from command $ fleetctl list-units)
$ docker exec -it wa_worker bash
[root@8f35d4e0aeef /]#
Next add your whatsapp credentials (you must use the yowsup library for registration: https://github.com/tgalal/yowsup/wiki/yowsup-cli-2.0#yowsup-cli-registration):
[root@8f35d4e0aeef /]# python $MOUNT_POINT/wa_worker/wa_worker/message_receiver/keystore.py
Usage:
python keystore.py wa_account wa_password
python keystore.py wa_account wa_password encrypt_key
Example
python keystore.py 5212288227733 ExamPlE/Wq+i/KEy912qwWqabJJ=
(Encrypt key will be taken from environment)
python keystore.py 5212288227733 ExamPlE/Wq+i/KEy912qwWqabJJ= MyEncrYptKey
(Encrypt key taken from param)
So you must add your whatsapp credentials, if SECRET_KEY environment var was defined you not need specify secret for encrypt this data.
- You can add more credentials for use it in case of banned whatsapp account, so you ensure that whatsapp message will always be delivered.
Specify email for support fail over email message sending:
[root@8f35d4e0aeef /]# python $MOUNT_POINT/wa_worker/wa_worker/message_receiver/mail.py
Usage:
python mail.py host user password admin
python mail.py host user password admin encrypt_key
Example:
python mail.py 172.16.16.16 fallout@mydomain.com xFGt3Swq support@devops.mydomain.com
(Encrypt key will be taken from environment)
python mail.py 172.16.16.16 fallout@mydomain.com xFGt3Swq support@devops.mydomain.com MyEncrYptKey
(Encrypt key taken from param)
This email account will be used only in case of failure delivering whatsapp message (because account banned or requested for new password by Whats App and no more whatsapp accounts configured)
###4) Start the daemon
Previously you had to comment lines in start.sh its time to uncomment the lines:
echo -e ' [*] Starting message receiver...\n';
python $MOUNT_POINT/wa_worker/wa_worker/message_receiver/start.py &
You can start it manual or you can save your changes, commit your project and leave that every happens automatically every once start your wa_worker service
###5) Send a message
You need a rabbitmq client that use the queue defined in MQ_SEND_MESSAGE_QUEUE environment var, the format of "body" is a json message:
{
"phones": ["5212287779788", "5212287779789"],
"mails": ["ivandavid77@gmail.com","dbarron@crediland.com.mx"],
"msg": "some message#13with new lines#13replaced"
}
If message has newlines it must be replaced for "#13" string.
You can use python "send_message.py" script for sending test messages:
[root@8f35d4e0aeef /]# python $MOUNT_POINT/wa_worker/wa_worker/message_receiver/utilities/send_message.py -h
usage: send_message.py [-h] [--phones PHONES [PHONES ...]]
[--mails MAILS [MAILS ...]] --msg MSG
optional arguments:
-h, --help show this help message and exit
--phones PHONES [PHONES ...]
phone list
--mails MAILS [MAILS ...]
mail list
--msg MSG "message to send"