Hi list, I''ve two servers running the same ldap service. A is primary server and B is simply a mirror, and accounts from server A are constantly synchronized with B. Now I don''t have a automatic failover solution, whenever server A goes down, I have to point the domain''s A record to server B and wait for the DNS to get updated. Takes quite a lot of time but the ldap service is crucial to the system and shouldn''t be down for longer than a few minutes. I''ve heard about Dyndns? They provide automatic failover if one server does down, what I want to know is if I can do the same using Bind? And if it''s possible to do it automatically? Without human intervention. I don''t want a round-robin solution, I want a automatic switchover if primary box goes down. Regards Per Qvindesland
Per Qvindesland wrote:> Hi list, > > > I''ve two servers running the same ldap service. A is primary server and B is > simply a mirror, and accounts from server A are constantly synchronized with > B. > > Now I don''t have a automatic failover solution, whenever server A goes down, > I have to point the domain''s A record to server B and wait for the DNS to > get updated. Takes quite a lot of time but the ldap service is crucial to > the system and shouldn''t be down for longer than a few minutes. > > I''ve heard about Dyndns? They provide automatic failover if one server does > down, what I want to know is if I can do the same using Bind? And if it''s > possible to do it automatically? Without human intervention. > > I don''t want a round-robin solution, I want a automatic switchover if > primary box goes down. > > Regards > Per Qvindesland > > > _______________________________________________ > CentOS mailing list > CentOS at centos.org > http://lists.centos.org/mailman/listinfo/centos >You could try something like this: http://dag.wieers.com/howto/bits/bind-ddns.php Bgrds, Finnur
Hi, If it is a mission critical system, check out http://www.linux-ha.org/ I have that managing a cluster of servers for HA SQL. In your case you may get away with just using a floating IP. Have the DNS pointing to that. Then no matter what server the IP is on the DNS doesn''t need to be updated. the fail over times in this type of setup are relativley low, maybe <1m depending on how you set it up .. Another Idea anyway. Darrin 2009/2/20 Per Qvindesland <per at norhex.com>> Hi list, > > > I''ve two servers running the same ldap service. A is primary server and B > is > simply a mirror, and accounts from server A are constantly synchronized > with > B. > > Now I don''t have a automatic failover solution, whenever server A goes > down, > I have to point the domain''s A record to server B and wait for the DNS to > get updated. Takes quite a lot of time but the ldap service is crucial to > the system and shouldn''t be down for longer than a few minutes. > > I''ve heard about Dyndns? They provide automatic failover if one server does > down, what I want to know is if I can do the same using Bind? And if it''s > possible to do it automatically? Without human intervention. > > I don''t want a round-robin solution, I want a automatic switchover if > primary box goes down. > > Regards > Per Qvindesland > > > _______________________________________________ > CentOS mailing list > CentOS at centos.org > http://lists.centos.org/mailman/listinfo/centos >-- Darrin Khan medavian at gmail.com Bill Watterson - "There is not enough time to do all the nothing we want to do." -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.centos.org/pipermail/centos/attachments/20090220/5923e10e/attachment.html
On Fri, Feb 20, 2009 at 1:29 PM, Per Qvindesland <per at norhex.com> wrote:> > I''ve two servers running the same ldap service. A is primary server and B is > simply a mirror, and accounts from server A are constantly synchronized with > B. > > Now I don''t have a automatic failover solution, whenever server A goes down, > I have to point the domain''s A record to server B and wait for the DNS to > get updated. Takes quite a lot of time but the ldap service is crucial to > the system and shouldn''t be down for longer than a few minutes. > > I''ve heard about Dyndns? They provide automatic failover if one server does > down, what I want to know is if I can do the same using Bind? And if it''s > possible to do it automatically? Without human intervention. > > I don''t want a round-robin solution, I want a automatic switchover if > primary box goes down.I also recommend you use a service IP address that can be migrated between your 2 servers. You can use heartbeat to automate that or do that manually as you are doing now. But moving a service IP is much simpler and faster. Stay away from using DNS to manage the failover. As you said is can take a while. Regards, Tim -- Tim Verhoeven - tim.verhoeven.be at gmail.com - 0479 / 88 11 83 Hoping the problem magically goes away by ignoring it is the "microsoft approach to programming" and should never be allowed. (Linus Torvalds)
Hi All Thanks for all your answers, I agree it would be better with heartbeat then to mock around with dns and a very slow update time. Regards Per Qvindesland On 2/20/09 2:08 PM, "Tim Verhoeven" <tim.verhoeven.be at gmail.com> wrote:> On Fri, Feb 20, 2009 at 1:29 PM, Per Qvindesland <per at norhex.com> wrote: >> >> I''ve two servers running the same ldap service. A is primary server and B is >> simply a mirror, and accounts from server A are constantly synchronized with >> B. >> >> Now I don''t have a automatic failover solution, whenever server A goes down, >> I have to point the domain''s A record to server B and wait for the DNS to >> get updated. Takes quite a lot of time but the ldap service is crucial to >> the system and shouldn''t be down for longer than a few minutes. >> >> I''ve heard about Dyndns? They provide automatic failover if one server does >> down, what I want to know is if I can do the same using Bind? And if it''s >> possible to do it automatically? Without human intervention. >> >> I don''t want a round-robin solution, I want a automatic switchover if >> primary box goes down. > > I also recommend you use a service IP address that can be migrated > between your 2 servers. You can use heartbeat to automate that or do > that manually as you are doing now. But moving a service IP is much > simpler and faster. Stay away from using DNS to manage the failover. > As you said is can take a while. > > Regards, > Tim
On Fri, 2009-02-20 at 15:36 +0100, Per Qvindesland wrote:> Hi All > > Thanks for all your answers, I agree it would be better with heartbeat then > to mock around with dns and a very slow update time. > > Regards > Per Qvindesland >Another benefit is that failover occurs much more quickly when using a floating IP managed with heartbeat compared to using multiple server listings in your clients dns and ldap config files. The default timeout for EACH lookup is killer (if not using client side caching) when the first server is down and the responses must come from the second one in the list.