Re: (ITS#6579) syncrepl lost in refreshAndPersist mode

[quanah@zimbra.com]

--On Monday, June 21, 2010 7:42 AM +0000 haller@efrei.fr wrote:

> Full_Name: Nicolas Haller
> Version: 2.4.22
> OS: FreeBSD 8.0
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (2001:7a8:1:15::143)
>
>
> Hi,
>
> I have a master LDAP server and two slaves. The slaves use syncrepl in
> refreshAndPersist mode.
>
> it's working until I see the message "connection_read(128): no
> connection!" in a slave log file. Here, I must restart the slave slapd to
> force the slave to be syncronized.
>
> I post the syncrepl slave conf:
>
> syncrepl rid=1
>   provider=ldap://ldap-master.blah.net
>   type=refreshAndPersist
>   interval=00:00:05:00
>   retry="30 +"
>   searchbase="dc=blah,dc=net"
>   schemachecking=on
>   bindmethod=simple
>   binddn="cn=root,dc=blah,dc=net"
>   credentials=blahblah

It sounds like you have a device between the master and the replicas that 
cuts the connection after some period of time.  If the operating system 
supports it, you can set a tcp keepalive in the sycnrepl configuration 
block in 2.4.22.

--Quanah

--

Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra ::  the leader in open source messaging and collaboration