On 07/20/2015 08:01 AM, Dave Page wrote:
>> Suggestion to resolve the issue:
>>
>> get the max value of ID from table before insert query and run the
>> sequence till we get the max(ID)+1 and then we execute the insert statement
>> with this new value as ID.
>
> We're not doing that - it'll be a performance killer. The user needs
> to not muck about with their sequences.
Yeah, pgagent doesn't need to cope with sequence failures. That's the
DBA's job to fix.
>
> The real questions are:
>
> 1) Why doesn't the insert error appear in the pgAgent log?
>
> 2) Can you reproduce Josh's issue where the first step in a job runs,
> but the second doesn't (without touching the sequences)?
Particularly, *after* the sequence errors are fixed, do the jobs now run
normally?
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com
