LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

[lvs-users] dovecot gone berserk w/LVS

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: [lvs-users] dovecot gone berserk w/LVS
From: Gerry Reno <greno@xxxxxxxxxxx>
Date: Thu, 24 Apr 2008 03:12:58 -0400
My LVS-DR has been working well.  But lately I needed some mailserver 
redundancy so I created another machine, put the mailstore on NFS, and 
the userdb in RDBMS and then created entries in my keepalived.conf for 
25, 110, and 143.  So I bring up just one mailserver to test behind the 
directors and postfix seems ok but dovecot is going crazy. 

LVS:          LVS-DR
VIP:          192.168.1.240
Directors:    192.168.1.150 and 151
Mailserver:   192.168.1.220

Here's what I'm seeing in the maillog:
Apr 24 02:56:27 grp-01-60-50 dovecot: imap-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:27 grp-01-60-50 dovecot: pop3-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:27 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3289
Apr 24 02:56:27 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3288
Apr 24 02:56:39 grp-01-60-50 dovecot: imap-login: Disconnected: 
rip=::ffff:192.168.1.150, lip=::ffff:192.168.1.220
Apr 24 02:56:39 grp-01-60-50 dovecot: pop3-login: Disconnected: 
rip=::ffff:192.168.1.150, lip=::ffff:192.168.1.220
Apr 24 02:56:39 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3292
Apr 24 02:56:40 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3293
Apr 24 02:56:42 grp-01-60-50 dovecot: imap-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:42 grp-01-60-50 dovecot: pop3-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:42 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3295
Apr 24 02:56:42 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3294
Apr 24 02:56:54 grp-01-60-50 dovecot: imap-login: Disconnected: 
rip=::ffff:192.168.1.150, lip=::ffff:192.168.1.220
Apr 24 02:56:54 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3358
Apr 24 02:56:55 grp-01-60-50 dovecot: pop3-login: Disconnected: 
rip=::ffff:192.168.1.150, lip=::ffff:192.168.1.220
Apr 24 02:56:55 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3359
Apr 24 02:56:57 grp-01-60-50 dovecot: imap-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:57 grp-01-60-50 dovecot: pop3-login: Disconnected: 
rip=::ffff:192.168.1.151, lip=::ffff:192.168.1.220
Apr 24 02:56:58 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3361
Apr 24 02:56:58 grp-01-60-50 dovecot: auth(default): new auth 
connection: pid=3360


Here is the master director log:
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:25] failed !!!
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:25] from VS [192.168.1.240:25]
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:143] failed !!!
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:143] from VS [192.168.1.240:143]
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:110] failed !!!
Apr 24 02:44:09 grp-01-00-50 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:110] from VS [192.168.1.240:110]
Apr 24 02:44:10 grp-01-00-50 Keepalived_vrrp: VRRP_Instance(VI_1) 
Transition to MASTER STATE
Apr 24 02:44:12 grp-01-00-50 Keepalived_vrrp: VRRP_Instance(VI_1) 
Entering MASTER STATE
Apr 24 02:44:12 grp-01-00-50 Keepalived_vrrp: VRRP_Instance(VI_1) 
setting protocol VIPs.
Apr 24 02:44:12 grp-01-00-50 avahi-daemon[2126]: Registering new address 
record for 192.168.1.240 on eth0.IPv4.
Apr 24 02:44:13 grp-01-00-50 Keepalived_healthcheckers: Netlink 
reflector reports IP 192.168.1.240 added
Apr 24 02:44:13 grp-01-00-50 Keepalived_vrrp: VRRP_Instance(VI_1) 
Sending gratuitous ARPs on eth0 for 192.168.1.240
Apr 24 02:44:13 grp-01-00-50 Keepalived_vrrp: Netlink: skipping nl_cmd 
msg...

Here is the backup director log:
Apr 24 02:45:09 grp-01-00-51 Keepalived_vrrp: VRRP_Instance(VI_1) 
Entering BACKUP STATE
Apr 24 02:45:09 grp-01-00-51 Keepalived_vrrp: VRRP sockpool: 
[ifindex(2), proto(112), fd(8,9)]
Apr 24 02:45:09 grp-01-00-51 kernel: IPVS: sync thread started: state = 
BACKUP, mcast_ifn = eth0, syncid = 25
Apr 24 02:45:09 grp-01-00-51 root: /etc/keepalived/manage_ip_lvs_dr  
(caller: keepalived): ip addr add 192.168.1.240/32 dev lo brd + scope host
Apr 24 02:45:09 grp-01-00-51 root: /etc/keepalived/manage_ip_lvs_dr  
(caller: keepalived): RTNETLINK answers: File exists
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:25] failed !!!
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:25] from VS [192.168.1.240:25]
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:143] failed !!!
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:143] from VS [192.168.1.240:143]
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.200:8080] failed !!!
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: Disabling 
service [192.168.1.200:8080] from VS [192.168.1.240:8080]
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.201:8080] failed !!!
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: Disabling 
service [192.168.1.201:8080] from VS [192.168.1.240:8080]
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:110] failed !!!
Apr 24 02:45:09 grp-01-00-51 Keepalived_healthcheckers: Disabling 
service [192.168.1.220:110] from VS [192.168.1.240:110]
Apr 24 02:45:54 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:143] success.
Apr 24 02:45:54 grp-01-00-51 Keepalived_healthcheckers: Enabling service 
[192.168.1.220:143] to VS [192.168.1.240:143]
Apr 24 02:45:54 grp-01-00-51 Keepalived_healthcheckers: TCP connection 
to [192.168.1.220:110] success.
Apr 24 02:45:54 grp-01-00-51 Keepalived_healthcheckers: Enabling service 
[192.168.1.220:110] to VS [192.168.1.240:110]


In keepalived.conf:
virtual_server 192.168.1.240 143 {
        delay_loop 15
        lb_algo rr
        lb_kind DR
        persistence_timeout 86400
        protocol TCP

        real_server 192.168.1.220 143 {
                weight 1
                inhibit_on_failure
                TCP_CHECK {
                        connect_port 143
                        connect_timeout 6
                        nb_get_retry 3
                        delay_before_retry 1
                }
        }
}

# same for 110


Anybody familiar with using dovecot behind the directors?  Is this an 
auth problem?  It says 'new auth connection' so I wouldn't think so.  
But the connection appears to just die.

????

Gerry



<Prev in Thread] Current Thread [Next in Thread>