Hello Frederic,
with this configuration you probably have some deadlocks in packetfence.log.
I did a setup with Haproxy and this is the config i used:
listen galera 172.20.135.3:3306
balance source
maxconn 2000
mode tcp
option tcplog
option tcpka
option mysql-check user haproxy
server MySQL1 172.20.135.1:3306 check weight 1
server MySQL2 172.20.135.2:3306 check weight 1 backup
listen galera-radius 172.20.135.3:3307
#balance roundrobin
balance source
mode tcp
maxconn 2000
rate-limit sessions 2000
option tcplog
option tcpka
option mysql-check user haproxy
server MySQL1 172.20.135.1:3306 check weight 1 backup
server MySQL2 172.20.135.2:3306 check weight 1
So port 3306 has been set in pf.conf and port 3307 in sql.conf (radius).
Also be carefull of pfdhcplistener process, only one instance can run on your
cluster.
I am also working on actif/actif setup for packetfence.
Regards
Fabrice
Le Mardi 25 Novembre 2014 04:21 EST, Frederic Hermann <[email protected]> a
écrit:
> Hi List,
>
>
> We are currently using packtfence (4.3.0) on debian 7, with a distinct mysql
> server (actually, a percona mysql cluster).
>
>
> We would like to setup haproxy on the pf server, to have dynamic failover,
> and eventually load-balancing between the different nodes of the cluster.
>
>
> For the moment we have the following issue : pf seems to loose the connection
> to the DB very often, which leads to unexpected behavior.
> The logs are full of 'ERRNO 2006' errors, in the different pf modules.
>
>
> I suspect something is wrong in our haproxy configuration, as pf works well
> fi connected directly to one of the cluster node.
>
>
> So I would like to know if someone here is working with such a configuration,
> and could share some configuration advices for packetfence and haproxy setup.
>
>
> Our haproxy configuration is as follow:
>
>
> * On each mysql node, a small xinetd service run on port 9200 to allow
> haproxy to check the cluster status using HTTP
> (This is to make sure that haproxy connect to a node actually connected to
> the cluster, and not to a node where mysql is running, but not in cluster.)
>
>
> * On pf, haproxy is configured as such:
>
>
>
>
>
> defaults
> mode tcp
> log global
> option tcplog
> option dontlognull
> retries 3
> option redispatch
> # option http-server-close
> # option forwardfor except 127.0.0.0/8
> maxconn 2000
> contimeout 5000
> clitimeout 50000
> srvtimeout 50000
>
>
> # backend app
> listen mysql-cluster 127.0.0.1:3306
> mode tcp
> balance source
> option httpchk
> server node1 192.168.4.11:3306 check port 9200 inter 3000 rise 2 fall 1
> server node2 192.168.4.12:3306 check port 9200 inter 3000 rise 2 fall 1
> server node3 192.168.4.13:3306 check port 9200 inter 3000 rise 2 fall 1
>
>
>
>
>
>
> Cheers,
>
>
> --
> Frederic Hermann
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk
_______________________________________________
PacketFence-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/packetfence-users
