[Pgpool-general] [PgPool-II] Node degenerating

[Tatsuo Ishii]

I was unable to reproduce your problem. What I did was:

- create 3 database clusters
- run the same query (global3=# VACUUM ANALYZEmetadata."RC_Metadata";)
  via psql->pgpool

No degeneration occurred. Just pgpool reported syntax error.

Is there any specific condition to reproduce your problem?
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp

> Hello
> I've got three nodes in pgpool-II:
> sO21 - id:0
> sO22 - id:1
> db00 - id:2
> replication_mode = true
> load_balance_mode = true
> replication_stop_on_mismatch = true
> failover_if_affected_tuples_mismatch = true
> 
> I'm getting an error after somebody of developers enters wrong syntax query:
> 
> 
>    1.  2011-06-23 15:24:26 LOG:   pid 13894: SimpleQuery: Unable TO parse
>    the query: global3=# VACUUM ANALYZEmetadata."RC_Metadata";
>    2.  2011-06-23 15:24:26 LOG:   pid 13894: DB node id: 0 backend pid:
>    13906 STATEMENT: BEGIN
>    3.  2011-06-23 15:24:26 LOG:   pid 13894: DB node id: 1 backend pid:
>    13788 STATEMENT: BEGIN
>    4.  2011-06-23 15:24:26 LOG:   pid 13894: DB node id: 0 backend pid:
>    13906 STATEMENT: global3=# VACUUMANALYZE metadata."RC_Metadata";
>    5.  2011-06-23 15:24:26 LOG:   pid 13894: pool_send_and_wait: Error
> OR notice
>    message FROM backend: : DB node id: 0 backend pid: 13906 STATEMENT:
>    global3=# vacu
>    6.  um ANALYZE metadata."RC_Metadata"; message: syntax error AT OR near
>    "global3"
> 
> 
> As you can see somebody copy-pasted vacuum analyze query with "database=#
> ...".
> After this pgpool degrades node with id=2. Why is this happening?
> I know that:
> 
> *"you should not execute VACUUM via pgpool. Connect*>>>* PostgreSQL
> directry and execute VACUUM".*
> 
> But it is the second problem when syntax error causes degenerating of
> the second node.
> 
> 
> Here is the log from pgpool-II:
> http://pastebin.com/N3uKAG7k
> 
> I will appreciate any help.
> 
> PS: Tatsuo, maybe you have some scripts with test queries to check if
> replication works correctly after online recovery?
> 
> -- 
> Best regards,
> Koldaev Anton