On 9/23/2010 8:46 AM, Steve Singer wrote: > On 10-09-23 09:20 AM, Karl Denninger wrote: >> On 9/23/2010 7:56 AM, Brad Nicholson wrote: >>> On 10-09-22 08:01 PM, Karl Denninger wrote: >>>> If you use Slony, expect it to lose the replication status. >>>> >>>> I attempted the following: >>>> >>>> 1. Master and slaves on 8.4. >>>> >>>> 2. Upgrade one slave to 9.0. Shut it down, used pg_upgrade to >>>> perform >>>> the upgrade. >>>> >>>> 3. Restarted the slave. >>>> >>>> Slony appeared to come up, but said it was syncing only TWO tables >>>> (out >>>> of the 33 in the working set!) Of course that didn't work very >>>> well at >>>> all..... >>>> > > A few things to confirm > > 1. You were running 2.0.4 before the upgrade as well Yes. > 2. The slony binaries your running with on 9.0 were compiled against > 9.0. In particular the slony shared library Yes. > 3. If you do select * from _clustername.sl_table and compare the > tab_reloid values with what is in pg_class do they match? Too late. I dropped the slave and re-inserted it, and it has re-synced. > Ie > > select c.relname, t.tab_relname FROM pg_class c, _clustername.sl_table > t where c.oid=t.tab_reloid > > you should see matches if not you need to run REPAIR CONFIG on the > slave (event node= slave id, only on set to slave id)
-- Karl
<<attachment: karl.vcf>>
_______________________________________________ Slony1-general mailing list [email protected] http://lists.slony.info/mailman/listinfo/slony1-general
