Threads in PostgreSQL

On Mon, Dec 21, 2015 at 11:51 AM, sri harsha <sriharsha9992@gmail.com>
wrote:

Hi,

Is it possible to use threads in Postgresql ?? I am using threads in my
foreign data wrapper and i get the following error when i use the threads .

*ERROR: stack depth limit exceeded*
*HINT: Increase the configuration parameter "max_stack_depth" (currently
2048kB), after ensuring the platform's stack depth limit is adequate.*

PostgreSQL is a process-per-backend model.

Can you elaborate on what you are using the threads for?

On 12/20/2015 10:21 PM, sri harsha wrote:

Is it possible to use threads in Postgresql ??

multi-threading is not used anywhere in postgresql, and is not at all a
good idea.

--
john r pierce, recycling bits in santa cruz

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

On 12/21/15 01:24, Atri Sharma wrote:

On Mon, Dec 21, 2015, sri harsha <sriharsha9992@gmail.com> wrote:

I am using threads in my
foreign data wrapper and i get the following error when i use the threads .

*ERROR: stack depth limit exceeded*
*HINT: Increase the configuration parameter "max_stack_depth" (currently
2048kB), after ensuring the platform's stack depth limit is adequate.*

PostgreSQL is a process-per-backend model.

To elaborate on that, it is a process per backend and most of the code
in that process has not been written for safe execution by multiple
threads. If your code starts other threads that only do other things,
but only one ever touches PostgreSQL backend code, that can be ok.
This happens in PL/Java, but it uses several interrelated precautions
to make sure no thread ever enters PostgreSQL backend code unless every
other thread is known to be out.

Even if you are taking that precaution, if the backend code is entered
by a different thread than last executed there, the stack depth tests
may be made by comparing the last thread's stack base to the current
thread's stack pointer, which will naturally give you a bogus result.
There is some API in miscadmin.h for manipulating the backend's idea
of the stack base, but there be dragons.

I am far short of the knowledgeable voices here, but in case you don't
hear from them right away, that is at least how I understand the matter.

Chapman Flack

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Hi,

PostgreSQL is not using threads but it is possible to spawn thread in
your PostgreSQL extensions.
For example, I have used pool of threads in my IMCS extension.
But you need to build your extension with -pthread:

CUSTOM_COPT = -pthread

Also, please take in account that many PostgreSQL functions (even in/out
or comparison functions) are not reentrant: them are storing their state
in global variables.
So you will get race conditions if you are calling such functions from
multiple threads.

Concerning stack overflow, I think that the most probable reason is
trivial infinite recursion.
Did you inspect stack trace in debugger?