amplify-agent is not running after system reboot

[gpluess]

I'm using version 0.44-2~xenial.

To find out, what was going on, I did the following: systemctl status amplify-agent tells me this:

* amplify-agent.service - LSB: Stop/start nginx-amplify-agent
   Loaded: loaded (/etc/init.d/amplify-agent; bad; vendor preset: enabled)
   Active: failed (Result: exit-code) since Tue 2017-07-18 21:39:24 UTC; 7min ago
     Docs: man:systemd-sysv-generator(8)
  Process: 1147 ExecStart=/etc/init.d/amplify-agent start (code=exited, status=1/FAILURE)

I then looked at the /var/log/amplify-agent/agent.log log file and discovered this:

2017-07-18 21:39:24,227 [1300] MainThread Hostname: localhost is local
2017-07-18 21:39:24,270 [1300] MainThread configtest check
2017-07-18 21:39:24,444 [1300] MainThread failed POST "https://receiver.amplify.nginx.com:443/1.3/[my-api-key]/agent/", exception: "("bad handshake: Error([('', 'osrandom_rand_bytes', 'getrandom() initialization failed.')],)",)"
2017-07-18 21:39:24,444 [1300] MainThread [None] post https://receiver.amplify.nginx.com:443/1.3/[my-api-key]/agent/ 500 10 0 0.173

However, when I start the agent manually with systemctl start amplify-agent everything works fine.

[seanmavley]

Using systemctl doesn't work for me. Having same issue after restart.

Here's my error:

Nov 03 08:26:28 user amplify-agent[14545]:   File "/usr/lib/python2.7/dist-packages/amplify/agent/common/util/http.py", line 131, in post
Nov 03 08:26:28 user amplify-agent[14545]:     return self.make_request(url, 'post', data=data, timeout=timeout, json=json)
Nov 03 08:26:28 user amplify-agent[14545]:   File "/usr/lib/python2.7/dist-packages/amplify/agent/common/util/http.py", line 113, in make_request
Nov 03 08:26:28 user amplify-agent[14545]:     raise e
Nov 03 08:26:28 user amplify-agent[14545]: ConnectionError: HTTPSConnectionPool(host='receiver.amplify.nginx.com', port=443): Max retries exceeded w
Nov 03 08:26:28 user systemd[1]: amplify-agent.service: Control process exited, code=exited status=1

[dedm]

Thanks @seanmavley, it's a known issue, we need to start agent after the system has access to network. We will address this issue soon, please bear with us.

[dvershinin]

Same issue with CentOS 7. Why this hasn't been addressed in a year?!

After reboot:

● amplify-agent.service - LSB: start and stop nginx-amplify-agent daemon
   Loaded: loaded (/etc/rc.d/init.d/amplify-agent; bad; vendor preset: disabled)
   Active: failed (Result: resources) since Thu 2018-12-20 22:07:18 UTC; 10h ago
     Docs: man:systemd-sysv-generator(8)
  Process: 2497 ExecStart=/etc/rc.d/init.d/amplify-agent start (code=exited, status=0/SUCCESS)

Dec 20 22:07:18 ps.ethicalpets.co.uk amplify-agent[2497]: Could not connect to API via url https://receiver.amplify.nginx.com:443/1.4
Dec 20 22:07:18 ps.ethicalpets.co.uk amplify-agent[2497]: We will try to establish a connection once again in a minute.
Dec 20 22:07:18 ps.ethicalpets.co.uk amplify-agent[2497]: Could not connect to API via url https://receiver.amplify.nginx.com:443/1.4
Dec 20 22:07:18 ps.ethicalpets.co.uk amplify-agent[2497]: Giving up after three attempts...
Dec 20 22:07:18 ps.ethicalpets.co.uk runuser[2665]: pam_unix(runuser:session): session closed for user nginx
Dec 20 22:07:18 ps.ethicalpets.co.uk amplify-agent[2497]: [FAILED]
Dec 20 22:07:18 ps.ethicalpets.co.uk systemd[1]: PID file /var/run/amplify-agent/amplify-agent.pid not readable (yet?) after start.
Dec 20 22:07:18 ps.ethicalpets.co.uk systemd[1]: Failed to start LSB: start and stop nginx-amplify-agent daemon.
Dec 20 22:07:18 ps.ethicalpets.co.uk systemd[1]: Unit amplify-agent.service entered failed state.
Dec 20 22:07:18 ps.ethicalpets.co.uk systemd[1]: amplify-agent.service failed.

Always have to restart it manually after each reboot or get spammed with lots of emails about inactive agent.

[IWriteThings]

This issue is still plaguing me in 2019. Any fix for this yet?

[dvershinin]

I've discovered that if the agent fails to start with similiar agent.log messages as original poster's, that means not enough entropy for TLS connection at boot time.

To fix you may want to use rng-tools as outlined in this post of mine. In CentOS 7 it basically goes down to:

yum -y install rng-tools
systemctl enable rngd

And viola, agent works after each reboot :)

[BigIron5]

How has this not been fixed yet?

[dvershinin]

@BigIron5 see my previous comment. Since NGINX Amplify uses python requests library and it fails TLS connection if there's not enough entropy at boot time - it's not really Amplify agent's blame. Simply ensure the entropy is there before the agent starts.

[madhon]

@BigIron5 see my previous comment. Since NGINX Amplify uses python requests library and it fails TLS connection if there's not enough entropy at boot time - it's not really Amplify agent's blame. Simply ensure the entropy is there before the agent starts.

The same happens on for me on Ubuntu 18.0.4.3 using haveged to provide entropy

[adventurous51]

I have used a somewhat cludgy workaround to this problem but it works great for me. I use Monit to monitor my server. So, I told Monit to monitor Amplify but to not alert me if it fails and restart it if it is not started. It works like a charm. I monitor NGINX with monit so there is no need for me to be alerted if Amplify fails. The config file looks like this:

#########################################################################

amplify agent

check process amplify-agent with pidfile /var/run/amplify-agent/amplify-agent.pid noalert [email protected] start program = "/etc/init.d/amplify-agent start" if does not exist then restart

My server is ubuntu. Hope it can help someone.

Cheers!