[Solved] Chef-server fails to create new user

Chef Infra (archive)
1

I installed chef-server and ran the reconfigure command. But when I try to create the admin with the following command -
chef-server-ctl user-create USER_NAME FIRST_NAME LAST_NAME EMAIL 'PASSWORD' --filename FILE_NAME
I get this error -
ERROR: Error connecting to https://127.0.0.1/users/, retry 1/5
.
.
.
ERROR: Errno::ECONNRESET: Error connecting to https://127.0.0.1/users/ - Connection reset by peer - SSL_connect

2

May be try http in localhost

3

Hi ,

Can you paste here the knife.rb file and hostname of your chef-server machineā€¦I think its trying to connect local host rather than your chef-server IP.

4

Hi,

Iā€™ve just installed chef following the chef rally tutorials. So I havenā€™t explored knife yet. Iā€™ve done a standalone installation. After facing this error, I edited the /etc/opscode/chef-server.rb file (which was originally empty) and added the followingā€¦
server_name = "chef(DOT)controller(DOT)com"
lb[ā€˜vipā€™]=server_name
#api_fqdn = server_name
notification_email = ā€œNO-REPLY-CHEF@example(DOT)comā€
#######################
## nginx configuration #
########################
nginx[ā€˜urlā€™] = "https://#{server_name}"
nginx[ā€˜server_nameā€™] = server_name
nginx[ā€˜ssl_portā€™]=443

After this the IP address in the error changed to Error connecting to https://chef(DOT)controller(DOT)com/users/ - Connection reset by peer - SSL_connect

My hostname is chef(DOT)controller(DOT)com

I also modified the url in /etc/opscode/pivotal.rb hoping to fix the issue but it didnā€™t work either.
node_name "pivotal"
chef_server_url "https://chef(DOT)controller(DOT)com:443"
chef_server_root "https://chef(DOT)controller(DOT)com:443"
no_proxy "chef(DOT)controller(DOT)com"
client_key key.path
ssl_verify_mode :verify_none

5

Hi,

I would double check to make sure you have a resolvable hostname and FQDN (via the hostname and hostname -f commands, respectively).

I suspect that the issue is here:

lb[ā€˜vipā€™]=server_name

This setting is meant for specifying a virtual IP address. Unless you have a specific reason not to, itā€™s a safe bet that you should get rid of this and let Chef server default to listening on localhost. If youā€™re setting this to use "ā€œchef(DOT)controller(DOT)comā€ and itā€™s not working, I suspect that this goes back to the resolvability / fqdn issue.

In addition to confirming the output of hostname -f, it would be helpful to see the output for these commands:

  • chef-server-ctl status
  • chef-server-ctl tail nginx

Also, just to confirm - make sure youā€™re running chef-server-ctl reconfigure after applying any changes.

6

Hi,

Thanks for your response.

I checked my hostname as you asked and got this response - chef(dot)controller(dot)com
So I guess hostname is not an issue.
After that I ran chef-server-ctl status which gave this output:
run: bookshelf: (pid 985) 41s; run: log: (pid 984) 41s
run: nginx: (pid 3056) 2s; run: log: (pid 986) 41s
run: oc_bifrost: (pid 1007) 41s; run: log: (pid 1006) 41s
run: oc_id: (pid 1046) 41s; run: log: (pid 1008) 41s
run: opscode-erchef: (pid 998) 41s; run: log: (pid 997) 41s
run: opscode-expander: (pid 1005) 41s; run: log: (pid 1004) 41s
run: opscode-solr4: (pid 991) 41s; run: log: (pid 990) 41s
run: postgresql: (pid 1000) 41s; run: log: (pid 999) 41s
run: rabbitmq: (pid 982) 41s; run: log: (pid 981) 41s
run: redis_lb: (pid 994) 41s; run: log: (pid 993) 41s

BUTā€¦running this chef-server-ctl tail nginx returned an error.

==> /var/log/opscode/nginx/error.log <==
2017/07/26 21:00:19 [emerg] 3108#0: still could not bind()
2017/07/26 21:00:21 [emerg] 3111#0: bind() to 0.0.0.0:80 failed (98: Address already in use)

==> /var/log/opscode/nginx/rewrite-port-80.log <==
127.0.0.1 - - [22/Jul/2017:14:00:57 +0530] ā€œGET / HTTP/1.1ā€ 301 191 ā€œ-ā€ "curl/7.29.0"
127.0.0.1 - - [22/Jul/2017:14:01:07 +0530] ā€œGET / HTTP/1.1ā€ 301 191 ā€œ-ā€ ā€œcurl/7.29.0ā€

==> /var/log/opscode/nginx/current <==
2017-07-26_15:30:21.52031 nginx: [emerg] still could not bind()
2017-07-26_15:30:21.67545 nginx: [emerg] bind() to 0.0.0.0:80 failed (98: Address already in use)

So I ran netstat -tulpn | grep :80 and saw that httpd was listening on the port.
tcp6 0 0 :::80 :::* LISTEN 958/httpd

I stopped httpd and ran chef-server-ctl tail nginx to see if it worked but I got the same error as above. So I executed chef-server-ctl reconfigure hoping to fix it but it didnā€™t work.

I again ran netstat -tulpn | grep :80 to check what was occupying port 80 and got this result.
tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN 3995/nginx: master

Now Iā€™m really confused as to how nginx can listen on the same port and log that it failed to bind.

7

Thereā€™s definitely some additional tweaking involved if you want to run Chef server alongside other stuff, particularly another web server. The first thing Iā€™d do is get rid of the extra stuff youā€™ve added to chef-server.rb and pivotal.rb, and then do another chef-server-ctl reconfigure, just to get you back to a mostly default setup.

Assuming that you do want to run Chef server alongside Apache in the long run, what youā€™d need to do is change the ports that nginx uses:

nginx['non_ssl_port'] = 7001
nginx['ssl_port'] = 7002

7001 and 7002 are just example ports here. Whatever you select, be sure to check it against the ports that Chef uses.

1 Like
8

Thanks for all the help! I finally got it working.
This is how I did it.
I restored my VM to a cleaner state. Then stopped apache service, set my hostname and then ran the user create cmd. Worked like a charm!

9

Glad to hear itā€™s working! Marking this as solved.