Hi,
I am using openldap 2.4.23-12.el6.i686 for provider and openldap version 2.2.28-4AV1.1 for slave. My configuration is Provider with proxy consumer on one machin and slave on another machine.
I am getting error “CSN too old, ignoring” in logs(provide-proxycosumer config)
slap_queue_csn: queing 0x8d7fe85e 20110901223155.770109Z#000000#000#000000
slap_graduate_commit_csn: removing 0x8d805020 20110901223155.770109Z#000000#000#000000
slap_queue_csn: queing 0x8b9fb300 20110901223155.770109Z#000000#000#000000
syncprov_sendresp: cookie=rid=128,csn=20110901223155.770109Z#000000#000#000000
do_syncrep2: rid=128 cookie=rid=128,csn=20110901223155.770109Z#000000#000#000000
do_syncrep2: rid=128 CSN too old, ignoring 20110901223155.770109Z#000000#000#000000
slap_graduate_commit_csn: removing 0x8d80b9a8 20110901223155.770109Z#000000#000#000000
Do my both provider and slave machines needs times synched using NTP or lag of few seconds is allowed (1-2 seconds)?
Thanks Rupesh
Other than original query , I have additional queries about delta replication. If have more than one consumers, do I need to decalre seprate access logs for each of the replica ? If I want to use changelog insteadof acesslog, where can I find documentation ?
Regards Rupesh
On 9/1/11, Rupesh Thakkar rupeshthakkar@gmail.com wrote:
Do my both provider and slave machines needs times synched using NTP or lag of few seconds is allowed (1-2 seconds)?
Thanks Rupesh
I've seen a very similar issue (ContextCSNs being newer on my consumers), and I believe it was fixed in OpenLDAP (2.4.26). Do the “CSN too old, ignoring” error messages spam endlessly in your consumer logs?
As for your other questions. You only need a single accesslog, and the clocks on your consumer nodes do not need to be in synch with the Provider (in deltasync replication).
-Yuri
On Thu, Sep 1, 2011 at 12:39 PM, Rupesh Thakkar rupeshthakkar@gmail.comwrote:
Other than original query , I have additional queries about delta replication. If have more than one consumers, do I need to decalre seprate access logs for each of the replica ? If I want to use changelog insteadof acesslog, where can I find documentation ?
Regards Rupesh
On 9/1/11, Rupesh Thakkar rupeshthakkar@gmail.com wrote:
Do my both provider and slave machines needs times synched using NTP or
lag
of few seconds is allowed (1-2 seconds)?
Thanks Rupesh
--On Saturday, September 03, 2011 1:02 AM -0700 Yuri Bank yuribank@gmail.com wrote:
I've seen a very similar issue (ContextCSNs being newer on my consumers), and I believe it was fixed in OpenLDAP (2.4.26). Do the "CSN too old, ignoring" error messages spam endlessly in your consumer logs?
As for your other questions. You only need a single accesslog, and the clocks on your consumer nodes do not need to be in synch with the Provider (in deltasync replication).
Wrong. Your clocks should *always* be tightly in sync. Particularly with the advent of delta-sync MMR in the next release.
--Quanah
--
Quanah Gibson-Mount Sr. Member of Technical Staff Zimbra, Inc A Division of VMware, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration
openldap-technical@openldap.org
-
Quanah Gibson-Mount -
Rupesh Thakkar -
Yuri Bank