Richard Yen dba
Mon Sep 18 02:01:34 PDT 2006
Hi,

I've got a setup that looks like this:
Node 2 -> Node 4 -> Node 5

I recently ran test_slony_state-dbi.pl on my database and it returned  
the following issues:

> ISSUES FOUND:
>
>
> Node: 2 Confirmations not propagating from 2 to 4
> ================================================
> Confirmations not propagating quickly in sl_confirm -
>
> For origin node 2, receiver node 4, earliest propagated
> confirmation has age 06:16:00 > 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that confirmations are not  
> propagating?
>
>
> Node: 2 Confirmations not propagating from 2 to 5
> ================================================
> Confirmations not propagating quickly in sl_confirm -
>
> For origin node 2, receiver node 5, earliest propagated
> confirmation has age 06:16:00 > 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that confirmations are not  
> propagating?
>
>
> Node: 4 Confirmations not propagating from 4 to 5
> ================================================
> Confirmations not propagating quickly in sl_confirm -
>
> For origin node 4, receiver node 5, earliest propagated
> confirmation has age 06:17:00 > 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that confirmations are not  
> propagating?
>
>
> Node: 4 Events not propagating to node 4
> ================================================
> Events not propagating quickly in sl_event -
> For origin node 4, earliest propagated event of age 06:17:00 >  
> 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that events are not propagating?
>
>
> Node: 5 Confirmations not propagating from 5 to 4
> ================================================
> Confirmations not propagating quickly in sl_confirm -
>
> For origin node 5, receiver node 4, earliest propagated
> confirmation has age 06:17:00 > 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that confirmations are not  
> propagating?
>
>
> Node: 5 Events not propagating to node 5
> ================================================
> Events not propagating quickly in sl_event -
> For origin node 5, earliest propagated event of age 06:17:00 >  
> 00:30:00
>
> Are slons running for both nodes?
>
> Could listen paths be missing so that events are not propagating?


Slon daemons are running on all three nodes.  The listen paths are  
currently as follows:
db=# select * from _schemaname.sl_listen ;
li_origin | li_provider | li_receiver
-----------+-------------+-------------
          2 |           2 |           4
          2 |           4 |           5
          4 |           4 |           2
          4 |           4 |           5
          5 |           5 |           2
          5 |           5 |           4
(6 rows)


According to the slony-I listen paths documentation, it looks like  
(origin=5,provider=5,receiver=2) should actually be  
(origin=5,provider=4,receiver=2).  The current setup is what's  
automatically generated by the store_listen and subscribe_set commands.

Try as I might, I cannot seem to successfully use DROP LISTEN and  
STORE LISTEN to reshape my sl_listen table.  Slonik just happily  
processes everything, but when I go to look inside psql, nothing's  
changed.

I actually dropped the _schemaname schema, and rebuilt slony, and  
still getting the same output from the script.

Would anyone be able to help?
--Richard



More information about the Slony1-general mailing list