Closed VladimirZD closed 9 years ago
Can you quote it exactly?
If it is a problem with database connections then I suggest you install pgbouncer to manage them - that's what we do.
https://lists.okfn.org/pipermail/ckan-dev/2013-November/006183.html
But I'm not clear if it is that error.
On 19 March 2015 at 08:40, Vladimir Mašala notifications@github.com wrote:
Hi, we started testing portal with few more users... not too much (10+) but we started getting error (in ckan-apache.error.log) FATAL: sorry, too many clients already, referer:....
We don't have too many clients so i am wondering is there any know issues regarding this?
Any hints how to solve this?
thnx
— Reply to this email directly or view it on GitHub https://github.com/datagovuk/dgu-vagrant-puppet/issues/93.
@davidread
Hi,
additional info about this error.
We have made some configuration changes in apache ckan.conf file
#WSGIDaemonProcess singlethreaded #WSGIProcessGroup singlethreaded WSGIProcessGroup %{GLOBAL}
And we have set debug=false in ckan.ini.
After this changes we had to many clients error.
Reverting apache ckan.conf settings back to
WSGIDaemonProcess singlethreaded WSGIProcessGroup singlethreaded #WSGIProcessGroup %{GLOBAL}
resolved error.
Can you tell what is yours configuration in apache ckan.conf file, and how have you configured pgbouncer max_client_conn, default_pool_size?
Also i can't find limit for idle connection in pgbouncer how can i set this?
Thank you.
We tried playing about with WSGIDaemonProcess and WSGIProcessGroup but in the end we just removed these options from our apache config. You may have success with them though.
I've put our pgbouncer config here: https://gist.github.com/davidread/843e355f893a43547e82
And then in the ckan.ini point at pgbouncer:
sqlalchemy.url = postgresql://dgu:REDACTED@localhost:6432/ckan
BTW Please can you quote the exact error message you see in the log, for my benefit and for others.
On 19 March 2015 at 12:53, Tomislav Balaz notifications@github.com wrote:
@davidread https://github.com/davidread
Hi,
additional info about this error.
We have made some configuration changes in apache ckan.conf file
#WSGIDaemonProcess singlethreaded #WSGIProcessGroup singlethreaded WSGIProcessGroup %{GLOBAL}
And we have set debug=false in ckan.ini.
After this changes we had to many clients error.
Reverting apache ckan.conf settings back to
WSGIDaemonProcess singlethreaded WSGIProcessGroup singlethreaded #WSGIProcessGroup %{GLOBAL}
resolved error.
Can you tell what is yours configuration in apache ckan.conf file, and how have you configured pgbouncer max_client_conn, default_pool_size?
Also i can't find limit for idle connection in pgbouncer how can i set this?
Thank you.
— Reply to this email directly or view it on GitHub https://github.com/datagovuk/dgu-vagrant-puppet/issues/93#issuecomment-83544813 .
@davidread David thank you for answer, we had GO LIVE today and this problem happened very near publishing time so it was a bit interesting.. Luckly everything is working fine....
Traceback:
"
Traceback: (most recent call last):, referer: http://data.gov.hr/data/search?sort=metadata_modified+desc
File "/var/ckan/wsgi_app.py", line 9, in
Thank you for help
Hi, we started testing portal with few more users... not too much (10+) but we started getting error (in ckan-apache.error.log) FATAL: sorry, too many clients already, referer:....
We don't have too many clients so i am wondering is there any know issues regarding this?
Any hints how to solve this?
thnx