[Bucardo-general] Value is Out of Range for Smallint

[Ben Allen]

The following message is repeated many times once you start bucardo  
with bucardo_ctl. A few notes on the database setup, it's two test  
databases doing swap replication on the same host. Similar to my  
previous example I posted.

Oct  2 13:26:48 postgres[61133]: [2-1] ERROR:  value "61134" is out  
of range for type smallint
Oct  2 13:26:48 postgres[61133]: [2-2] STATEMENT:
Oct  2 13:26:48 postgres[61133]: [2-3]                   UPDATE  
bucardo.q
Oct  2 13:26:48 postgres[61133]: [2-4]                   SET     
aborted=timeofday()::timestamp, whydie=$1
Oct  2 13:26:48 postgres[61133]: [2-5]                   WHERE  sync  
= $2
Oct  2 13:26:48 postgres[61133]: [2-6]                   AND    pid = $3
Oct  2 13:26:48 postgres[61133]: [2-7]                   AND    ppid  
= $4
Oct  2 13:26:48 postgres[61133]: [2-8]                   AND     
targetdb = $5
Oct  2 13:26:48 postgres[61133]: [2-9]                   AND     
started IS NOT NULL
Oct  2 13:26:48 postgres[61133]: [2-10]                  AND    ended  
IS NULL
Oct  2 13:26:48 postgres[61133]: [2-11]                  AND     
aborted IS NULL
Oct  2 13:26:48 postgres[61133]: [2-12]
Oct  2 13:26:50 postgres[61149]: [2-1] ERROR:  value "61146" is out  
of range for type smallint
Oct  2 13:26:50 postgres[61149]: [2-2] STATEMENT:
Oct  2 13:26:50 postgres[61149]: [2-3]                            
UPDATE bucardo.q
Oct  2 13:26:50 postgres[61149]: [2-4]                            
SET    aborted=timeofday()::timestamp, whydie=$1
Oct  2 13:26:50 postgres[61149]: [2-5]                            
WHERE  sync=$2
Oct  2 13:26:50 postgres[61149]: [2-6]                            
AND    pid=$3
Oct  2 13:26:50 postgres[61149]: [2-7]                            
AND    ended IS NULL
Oct  2 13:26:50 postgres[61149]: [2-8]                            
AND    aborted IS NULL

When this is happening replication doesn't occur.
Few things I've found relating to this, this only happens on my AMD64  
system, but not my i386 system. Both are FreeBSD 6.2 with PostgreSQL  
8.2.5 installed from ports. For a quick fix, I just altered the "q"  
table by:

ALTER TABLE q ALTER ppid TYPE integer;
ALTER TABLE q ALTER pid TYPE integer;

After this fix replication occurs normally.

Regards,

Ben Allen


-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20071002/bfd7ccf8/attachment.html