Just upgrade from 5.00GA -> 5.01GA on Etch with ldap delay replicate problem

[Thanakorn]

I have 2 servers both are running Zimbra 5.01GA FOSS over Debian Etch

mail1 with everything except zimbra-proxy and mail2 with everything except zimbra-logger and zimbra-proxy.

mail2 ldap is relica of mail1

The upgrade went very well without any error on master and replica

, anyway on the replica both ldap_amavis_password and ldap_postfix_password doens't match the one on mail1 server so I have to change using zmlocalconfig command and do zmmtainit command to change ldap*.cf and amavisd.conf in the conf directory .

Well there is another weird problem I found. It seems that both server sync each other every hours instead of imediately.


and this is the zimbra.log from both server

mail1:/opt/zimbra/conf# cat /var/log/zimbra.log | grep slapd | more
Jan 17 07:59:44 10.1.100.2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server  
Jan 17 07:59:44 10.1.100.2 slapd[2962]: do_syncrepl: rid 100 retrying  
Jan 17 10:00:44 10.1.100.2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server  
Jan 17 10:00:44 10.1.100.2 slapd[2962]: do_syncrepl: rid 100 retrying  
Jan 17 12:01:45 10.1.100.2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server  
Jan 17 12:01:45 10.1.100.2 slapd[2962]: do_syncrepl: rid 100 retrying  
Jan 17 14:02:46 10.1.100.2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server  
Jan 17 14:02:46 10.1.100.2 slapd[2962]: do_syncrepl: rid 100 retrying  
mail1:/opt/zimbra/conf# 


mail2:/opt/zimbra/libexec# cat /var/log/zimbra.log | grep slapd | more
Jan 17 08:00:13 mail2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server 
Jan 17 08:00:13 mail2 slapd[2962]: do_syncrepl: rid 100 retrying 
Jan 17 10:01:14 mail2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server 
Jan 17 10:01:14 mail2 slapd[2962]: do_syncrepl: rid 100 retrying 
Jan 17 12:02:14 mail2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server 
Jan 17 12:02:14 mail2 slapd[2962]: do_syncrepl: rid 100 retrying 
Jan 17 14:03:16 mail2 slapd[2962]: do_syncrep2: rid 100 Can't contact LDAP server 
Jan 17 14:03:16 mail2 slapd[2962]: do_syncrepl: rid 100 retrying 
mail2:/opt/zimbra/libexec#

Both server connect to each other via vpn with 1Mb wan link and when the log said "do_syncrep2: rid 100 Can't contact LDAP server" then I can notice that it get update on the replica

I have another 2 servers and for authentication and GAL using openldap. That 2 servers I use syncrepl refreshOnly method instead of refreshandPersist that zimbra use and they sync each other every 5 minute without any problem.

I don't know is there any issue on using refreshandPersist on quite slow wan link or not.

Could someone suggest me about this? I know from seeing other post that sending hup signal to slapd or simply stop/start zimbra service should work but it should be better way to fix this.

[quanah]

Refresh and persist means that it keeps a persistent connection to the master. It sounds like your internet connection is flaky enough that it doesn't stay connected. You may want to add an "interval" statement to the slapd.conf.in file, or change to the refreshOnly mode.

[Thanakorn]

Today I added account using zmprov for 30 people and it change imediately. Maybe yesterday I add only 1 user.....so it's too less to notify the replica


Anyway it seem ok today and as you said, changing the replication mode to refreshOnly might be better for slow wan link.