unixsurfer / haproxyadmin Goto Github PK

import haproxyadmin
hap = haproxyadmin.haproxy.HAProxy(socket_file="/var/lib/haproxy/stats")
hap.clear_map(1) 
hap.show_map("/etc/haproxy/backend.map") # => ['']

Expected it to return []

Hello,

Do you plan to expose an REST API for this great plugin with flask for example ?

Thanks.

Are the 2 years of changes for this project going to be released on to PyPi. Or this project as is and needs to be cloned from master?

I ran into a problem setting the IP of servers, running with nbproc > 1. It seems HAProxy syncs the changes by itself over the processes, so that the outputs of the socket writes differ. I confirmed this with HAProxy v 1.7.11 and 1.8.6 on Centos 7:

Traceback (most recent call last):
  File "namerd-haproxy.py", line 270, in <module>
    main()
  File "namerd-haproxy.py", line 262, in main
    schedule["check"]()
  File "namerd-haproxy.py", line 194, in check_namerd
    haproxy_servers[target_haproxy_backend].setaddress(new_address) #, new_port)
  File "/vagrant/haproxyadmin/server.py", line 146, in setaddress
    return compare_values(values)
  File "/vagrant/haproxyadmin/utils.py", line 271, in compare_values
    raise IncosistentData(values)
haproxyadmin.exceptions.IncosistentData: Received different data per HAProxy process

The 'values' variable in 'utils/compare_values' had the following content:

[(1, "IP changed from '127.0.0.1' to '10.10.26.114' by 'stats socket command'"), (1, "no need to change the addr by 'stats socket command'"), (1, "no need to change the addr by 'stats socket command'"), (1, "no need to change the addr by 'stats socket command'")]

Used HAProxy config:

global
    log         127.0.0.1 local2
    chroot      /var/lib/haproxy
    pidfile     /var/run/haproxy.pid
    maxconn     4000
    user        haproxy
    group       haproxy
    daemon
    nbproc      4
    stats socket /run/haproxy/1.sock mode 0744 level admin process 1
    stats socket /run/haproxy/2.sock mode 0744 level admin process 2
    stats socket /run/haproxy/3.sock mode 0744 level admin process 3
    stats socket /run/haproxy/4.sock mode 0744 level admin process 4

defaults
    mode http

frontend  main
    bind *:5000
    default_backend             app

backend app
    balance     roundrobin
    server  app1 127.0.0.1:80 check

I tested the following on the command line to see if the IP propagates by itself through all HAProxy processes:

 [root@namerd-haproxy00 vagrant]# echo "show servers state" | socat /run/haproxy/1.sock -
 1
 # be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_fqdn srv_port
 7 app 1 app1 10.10.26.114 0 0 1 1 1574 8 2 0 6 0 0 0 - 80
 
 [root@namerd-haproxy00 vagrant]# echo "show servers state" | socat /run/haproxy/2.sock -
 1
 # be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_fqdn srv_port
 7 app 1 app1 10.10.26.114 0 0 1 1 1576 8 2 0 6 0 0 0 - 80
 
 [root@namerd-haproxy00 vagrant]# echo "set server app/app1 addr 10.10.10.10 port 3333" | socat /run/haproxy/1.sock -
 IP changed from '10.10.26.114' to '10.10.10.10', port changed from '80' to '3333' by 'stats socket command'
 
 [root@namerd-haproxy00 vagrant]# echo "show servers state" | socat /run/haproxy/2.sock -
 1
 # be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_fqdn srv_port
 7 app 1 app1 10.10.10.10 0 0 1 1 1594 7 2 0 6 0 0 0 - 3333
 
 [root@namerd-haproxy00 vagrant]# echo "show servers state" | socat /run/haproxy/3.sock -
 1
 # be_id be_name srv_id srv_name srv_addr srv_op_state srv_admin_state srv_uweight srv_iweight srv_time_since_last_change srv_check_status srv_check_result srv_check_health srv_check_state srv_agent_state bk_f_forced_id srv_f_forced_id srv_fqdn srv_port
 7 app 1 app1 10.10.10.10 0 0 1 1 1602 7 0 0 7 0 0 0 - 3333

https://github.com/unixsurfer/haproxyadmin/blob/master/haproxyadmin/internal.py line 89.

Currently, the connection is established via unix socket files. It would be nice if this library would also support connecting to HAProxy over TCP as well. :)

[EDIT]:
URLs might be a good way on how to define how it connects.

hap = haproxy.HAProxy('unix:///var/run/haproxy.sock')  # via unix socket
hap = haproxy.HAProxy('tcp://127.0.0.1:1937')  # via TCP

Hey Pavlos,

How are you doing? I'm investigating a problem with haproxyadmin, which sometimes fails to send commands through the socket. It turns out to be the function connected_socket() in utils.py, which has a hardcoded timeout of 0.1 seconds. It wouldn't be a problem if the function only opened a connection. But it does more than that; it also sends a show info command and expects a response. Sometimes we don't get a response in time, causing this function to fail.

I'm considering submitting a patch to make this timeout configurable. I would add a parameter to the constructor of the HAProxy class, similar to retry and retry_interval. This param could also be passed along to HAProxyProcess so it can be used when sending other commands (see internal.py file, line 82), for consistency.

Does it make sense?

The setstate function only worked when I added: "from haproxyadmin.server import *" on haproxy.py.
Probably this is not the best way to solve this problem.

Throws an exception on Python 2.7.9:

  File "/usr/local/lib/python2.7/dist-packages/haproxyadmin/haproxy.py", line 110, in __init__
    if is_unix_socket(_file) and connected_socket(_file):
  File "/usr/local/lib/python2.7/dist-packages/haproxyadmin/utils.py", line 179, in connected_socket
    except (ConnectionRefusedError, PermissionError, socket.timeout, OSError):
NameError: global name 'ConnectionRefusedError' is not defined

Packages version:

haproxyadmin (0.2.1)
six (1.11.0)

csv_data = self.command('show stat {} {} {}'.format(iid, obj_type, sid),
                                full_output=True)

temporary fix

csv_data = self.command('show stat {0} {1} {2}'.format(iid, obj_type, sid), full_output=True)

[root@host haproxyadmin]# python2.7 setup.py install
ERROR:root:Error parsing
Traceback (most recent call last):
File "/app/interpreters/python/2.7.11/lib/python2.7/site-packages/pbr/core.py", line 111, in pbr
attrs = util.cfg_to_args(path, dist.script_args)
File "/app/interpreters/python/2.7.11/lib/python2.7/site-packages/pbr/util.py", line 264, in cfg_to_args
wrap_commands(kwargs)
File "/app/interpreters/python/2.7.11/lib/python2.7/site-packages/pbr/util.py", line 566, in wrap_commands
cmdclass = ep.resolve()
File "/app/interpreters/python/2.7.11/lib/python2.7/site-packages/pkg_resources/init.py", line 2297, in resolve
module = import(self.module_name, fromlist=['name'], level=0)
ImportError: No module named scripts.setupcommands
error in setup command: Error parsing /root/haproxyadmin/setup.cfg: ImportError: No module named scripts.setupcommands
[root@cblb-e1a-01 haproxyadmin]# pip2.7 list
DEPRECATION: The default format will switch to columns in the future. You can use --format=(legacy|columns) (or define a format=(legacy|columns) in your pip.conf under the [list] section) to disable this warning.
appdynamics (4.2.3.0)
appdynamics-bindeps-linux-x64 (4.2.3.0)
appdynamics-proxysupport-linux-x64 (1.8.0.51)
awscli (1.10.52)
botocore (1.4.42)
colorama (0.3.7)
docutils (0.12)
futures (3.0.5)
jmespath (0.9.0)
meld3 (1.0.2)
pbr (1.10.0)
pip (9.0.1)
psutil (5.0.0)
pyasn1 (0.1.9)
python-dateutil (2.5.3)
rsa (3.4.2)
s3transfer (0.1.1)
setuptools (29.0.1)
six (1.10.0)
supervisor (3.1.3)
uWSGI (2.0.12)
wheel (0.29.0)

It returns ['<id> <key> <value>', ...]. Any reason not to return something like [(id, key, value), ...]?