Grails with Postgress ADMIN SHUTDOWN issue

[Vadim Kirilchuk]

Hello,

I have a grails application deployed on cloudfoundry and i`am facing next issue with it:

While trying to execute some update query i get "SQL Error: Already Closed" message.

If i look at my application log, i will see this:

2012-02-19 18:55:26,683 WARN  [hibernate.util.JDBCExceptionReporter] - SQL Error: 0, SQLState: 57P01
2012-02-19 18:55:26,683 ERROR [hibernate.util.JDBCExceptionReporter] - Batch entry 0 insert into TEXTEIN_EDITING_WORKFLOW (version, accepted, completed, cost, date_created, editor_id, last_updated, unit_id, id) values ('0', '0', '0', '0.0', '2012-02-19 18:55:22.790000 +00:00:00', '15', '2012-02-19 18:55:22.790000 +00:00:00', '39', '54') was aborted.  Call getNextException to see the cause.

2012-02-19 18:55:26,683 WARN  [hibernate.util.JDBCExceptionReporter] - SQL Error: 0, SQLState: 57P01
2012-02-19 18:55:26,683 ERROR [hibernate.util.JDBCExceptionReporter] - FATAL: terminating connection due to administrator command
2012-02-19 18:55:26,683 WARN  [hibernate.util.JDBCExceptionReporter] - SQL Error: 0, SQLState: 08006
2012-02-19 18:55:26,683 ERROR [hibernate.util.JDBCExceptionReporter] - An I/O error occured while sending to the backend.

I`m using postgres and error code "SQLState: 57P01" means ADMIN SHUTDOWN.
How is it possible? And what does it mean?

Could you shed me the light on this?

Thanks.

[Vadim Kirilchuk]

I changed the database to in-mem h2, and the problem has gone..

I will try to use postgres in my local environment and check if it is issue of cloudfoundry or postgress itself.

[pledbrook]

Does the error with PostgreSQL always occur? Or did it only happen once? Is it for a long running update? Or just a single record?

[Vadim Kirilchuk]

Hi, Peter,

I haven`t tried postgres on my local environment yet..

Does the error with PostgreSQL always occur? Or did it only happen once?

It occurs from time to time with 50 percent probability =)

Also i thought that the problem was in stale connections, but changing connection pool properties didn`t help.

Is it for a long running update?

No, it is just simple insert or update, usually - single record. But the update is inside a transaction.. And another important thing - i have something like this:

grails auto transaction start
do something with entity
save (but not flush and not commit)
call to email service which actually has "transactional=false"
send email (email send is successful)
done
grails auto end transaction

But, again, actually i`m not sure that the problem is in cloudfoundry or postgress.. I need to check this on my local environment.