We have only a few databases, but a lot of tables that are updated quite
frequently. E.g. our 'trumgr_group' db has 100 tables for unit_history,
and 100 tables for unit_stats.
Last week we performed a (sort-of) ETL to merge 4 million records for
22,000 'legacy' router units that were maintained in old infrastructure.
Our import went very well, and I simply did a redeployment of bucardo
from 'scratch' after we were done to set up everything fresh on the
replicant masters.
However, with all the new data, the syncs seem to be laboring a bit. Our
routers can 'talk' to either of our masters, and we're now getting
Conflicts. The conflicts get resolved - but bucardo will throw a 'could
not serialize', and the sync will go 'bad' for a while. It eventually
seems to get itself straightened out on its own, but for the interval
when it's doing so, we get inconsistent data reported in our application
that customer service uses to monitor and review individual router
statistics and data.
What I'm wondering is, could I break up the sync of those hundreds of
tables that are in one db into separate syncs for say each ten tables?
And more importantly - would there be any value in this - would having
separate syncs make the each sync iteration a smaller 'chunk' which
bucardo could chrew through more effectively, and resolve conflicts more
efficiently?
--
Paul Theodoropoulos
www.anastrophe.com
_______________________________________________
Bucardo-general mailing list
[email protected]
https://mail.endcrypt.com/mailman/listinfo/bucardo-general
