Thread: is there any backend timeout undocumented?
Hi! I have sended the message below to pgadmin-support but receive no answers... I hope you can help me on this... Is there any server timeout that is undocumented? I've issued a query like the one below and the server timed out after 180min (+/-). The query "construct_warehouse()" can last well above the 180min because it fills a table with millions of tuples... ---------------------------------------------------------------------------- spid=> vacuum full analyze ; select construct_warehouse() ; vacuum analyze ; NOTICE: Skipping "pg_group" --- only table or database owner can VACUUM it NOTICE: Skipping "pg_database" --- only table or database owner can VACUUM it NOTICE: Skipping "pg_shadow" --- only table or database owner can VACUUM it VACUUM server closed the connection unexpectedlyThis probably means the server terminated abnormallybefore or while processing therequest. The connection to the server was lost. Attempting reset: Failed. You are currently not connected to a database. !> \q ---------------------------------------------------------------------------- I've searched the archives for some documented timeout but nothing... I've searched the postgresql.conf file and nothing... Can anyone help me? Thanks in advance! Note: the first time I noticed a time out was using a JDBC driver and then I've tested in the pgsql to confirm it. -- o__ Bem haja, _.>/ _ NunoACHenriques (_) \(_) ~~~~~~~~~~~~~~~ http://students.fct.unl.pt/users/nuno/
NunoACHenriques <nach@fct.unl.pt> writes:
> Is there any server timeout that is undocumented?
No.
> spid=> vacuum full analyze ; select construct_warehouse() ; vacuum analyze ;
> NOTICE: Skipping "pg_group" --- only table or database owner can VACUUM it
> NOTICE: Skipping "pg_database" --- only table or database owner can VACUUM it
> NOTICE: Skipping "pg_shadow" --- only table or database owner can VACUUM it
> VACUUM
> server closed the connection unexpectedly
> This probably means the server terminated abnormally
> before or while processing the request.
This looks like a crash to me, not a timeout. Can you provide us with a
stack backtrace? Also, you'd better explain what construct_warehouse()
is doing.
regards, tom lane
On Thu, 23 May 2002 18:36:17 +0100 (WEST), NunoACHenriques <nach@fct.unl.pt> wrote: >server closed the connection unexpectedly > This probably means the server terminated abnormally > before or while processing the request. >The connection to the server was lost. Attempting reset: Failed. >You are currently not connected to a database. I've seen this before. In my case it was not a timeout, but a backend crash. What version are you running? Do you find anything useful in the log file? ServusManfred
On Thu, 23 May 2002 18:36:17 +0100 (WEST) "NunoACHenriques" <nach@fct.unl.pt> wrote: > Is there any server timeout that is undocumented? Looks more like a backend crash to me. Can you look for a core file in $PGDATA/base/xxx/ (where xxx is the OID of your database)? If you don't have debugging already enabled, try rebuilding PostgreSQL with debugging support (./configure --enable-debug, or "-g" CFLAGS), and then getting a backtrace with gdb. Also, posting the source of construct_warehouse() might be helpful. Cheers, Neil -- Neil Conway <neilconway@rogers.com> PGP Key ID: DB3C29FC
NunoACHenriques <nach@fct.unl.pt> writes: > I've issued a query like the one below and the server timed out after > 180min (+/-). The query "construct_warehouse()" can last well above the > 180min because it fills a table with millions of tuples... > > ---------------------------------------------------------------------------- > spid=> vacuum full analyze ; select construct_warehouse() ; vacuum analyze ; > NOTICE: Skipping "pg_group" --- only table or database owner can VACUUM it > NOTICE: Skipping "pg_database" --- only table or database owner can VACUUM it > NOTICE: Skipping "pg_shadow" --- only table or database owner can VACUUM it > VACUUM > server closed the connection unexpectedly > This probably means the server terminated abnormally > before or while processing the request. > The connection to the server was lost. Attempting reset: Failed. > You are currently not connected to a database. As the message says the backend is not "timing out"; it's terminating *abnormally*, What's doing the construct_warehouse() function?, It's written in C?, Could you send the backtrace from the core file? Regards, Manuel.