Whoa, I guess there's something I really don't understand about  
slony.  Perhaps you can answer some more fundamental questions for me?

As I understand it, the sync_interval parameter is the frequency at  
which a SYNC event is sent from the local machine to all other  
machines in the cluster.  So if sync_interval is 3000, I'll be  
sending SYNC events every 3 seconds.

But what does this say about data changes, like INSERTs and UPDATEs?   
Do these wait until the SYNC event is sent, or are they processed ASAP?

Also, I wonder if I'm overly-concerned about lag_time and  
lag_num_events; I'm running a 3-node system where all SELECTS go to  
the 2 subscribers and all data changes go to the provider.   
Naturally, I'd want all the data to be synced ASAP, so that users  
don't get confused.  I'm talking about data like contact info,  
timestamps, etc.  Right now, I get warning messages when we hit 40  
lag_num_events or 60 sec lag_time, whichever comes first.

Any insights?

Thanks for the help!
--Richard




On Sep 8, 2006, at 8:12 AM, Vivek Khera wrote:

>
> On Sep 7, 2006, at 2:57 PM, Richard Yen wrote:
>
>> I've been experiencing consistent lag on my subscriber nodes, and no
>> amount of VACUUMing (full, and just analyze) has been able to make
>> the lag go away.
>
> well, depending on how often you sync, the lag of 10 events could  
> be normal.
>
> i personally don't bother worrying until the lag_time gets to be >3  
> minutes which for me is hundreds of events.
>
> _______________________________________________
> Slony1-general mailing list
> Slony1-general at gborg.postgresql.org
> http://gborg.postgresql.org/mailman/listinfo/slony1-general