Way to increase number of clients?

[jerlyn06]

Hello all, 

We're on the enterprise version TACTIC 4.3.0.v02, and we're working on building integration tools into our DCC apps. As such, once a file gets 'checked in', it would launch a Python script that changes the status to submitted_for_review. However, very frequently, especially when multiple users are checking in, we receive the following error: 

Code:

Traceback (most recent call last):
File "tactic_update.py", line 58, in <module>
File "tactic_update.py", line 37, in update_tactic
File "tactic_client_lib\tactic_server_stub.py", line 816, in finish

File "xmlrpclib.py", line 1243, in __call__
return self.__send(self.__name, args)
File "xmlrpclib.py", line 1602, in __request
verbose=self.__verbose
File "xmlrpclib.py", line 1283, in request
return self.single_request(host, handler, request_body, verbose)
File "xmlrpclib.py", line 1316, in single_request
return self.parse_response(response)
File "xmlrpclib.py", line 1493, in parse_response
return u.close()
File "xmlrpclib.py", line 800, in close
raise Fault(**self._stack[0])
xmlrpclib.Fault: <Fault 1: 'FATAL:  sorry, too many clients already\n'>
[4528] Failed to execute script tactic_update[/code]


Any way to increase the number of clients?

[listy]

Hi!
You should incrase max_connections of Your pqsql server.
Hope this helps.
Bye

[remkonoteboom]

Another possible cause of this issue is the config setting

<database>
...
<pool_max_connections>0</pool_max_connections>
...
</database>

For most installations, this setting should be 0. What this does is ensure that every request will open only the connection it uses and will close all of them when request exits. If it is not 0, then what will happen is that TACTIC will keep connections open across requests. This will occur for each database connected and for each thread of every python process. This can grow quite large if you have a large number of processes and /or projects. I would only recommend using this if latency to the database connections are a significant factor (we have observed this is Azure managed databases where connections to the database are unacceptably large and forcing a pool requirement)

As listy mentioned, you can increase the number of database connections for postgresql, but you may have to increase shared memory and semaphores. For Linux systems, we have used something like:

echo "134217728" >/proc/sys/kernel/shmmax
echo "134217728" >/proc/sys/kernel/shmall

Place an entry in /etc/sysctl.conf like,

kernel.shmmax=134217728
kernel.shmall=134217728

You can find more details on the Postgresql website.

[jerlyn06]

Thank you @listy and @remkonoteboom. I'll take a look at the postgresql server.

[remkonoteboom]

Just for completion, my resource for how to do this is:

https://www.postgresql.org/docs/9.4/kern...urces.html