The pool number is set to 10 .
I tried also with 10 worker threads
For what can i see there is something wrong in my app.
The example app and a wizard app are correctly processed.
Also the static pages like .html are fast rendered instead the pages
rendered by the controller are very slow
[Fri Oct 21 18:02:29.083973 2016] [mpm_winnt:notice] [pid 620:tid 320]
AH00354: Child: Starting 250 worker threads.
This is a ridiculous high number of threads! ;-)
If 10 is the magic number try looking at the pool number in your
appconfig.ini.
If you need high concurrency try avoiding threads
Hi massimo and thanks fot the hint
I tried with rocket from cmd line and this is result from Apache ab command
line.
I got the same result i have with APACHE WSGI as you can see below
*MY APP WITH ROCKET *
This is the command launch
C:\web2py>python web2py.py
*No handlers could be found for
Something is very wrong here and I do not know what it is. Why are you
suing apache with WSCGI. This is not a recommended configuration. We do not
support Apache any more and we never supported WSCGI. Can you try rocket
and nginx?
On Monday, 24 October 2016 10:47:49 UTC-5, piero crisci wrote:
I update the version and i tried also on a WINDOWS 7 with the same
configuration.
Here is the difference beetwen the WINDOWS SERVER 2012R2 and WINDOWS 7 with
the same configuration
*APACHE*
*Server version: Apache/2.4.23 (Win64)*
*Server built: Oct 1 2016 08:28:43*
*Distributed by: The
I just checked error logs and i got this:
[Fri Oct 21 18:02:28.797956 2016] [wsgi:warn] [pid 4676:tid 436] mod_wsgi:
Compiled for Python/2.7.9+.
[Fri Oct 21 18:02:28.798955 2016] [wsgi:warn] [pid 4676:tid 436] mod_wsgi:
Runtime using Python/2.7.9.
[Fri Oct 21 18:02:28.799957 2016]
perhaps its easier to give the error log rather than just the
configuration, learn from a lot of discussion in this forum apache is not
recommended
best regards,
stifan
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py (Source code)
-
Further update:
Apache did finally crash, with the following message:
Server ran out of threads to serve requests. Consider raising the
ThreadsPerChild setting
I've bumped that up to 500, and there are less than that number of users on
the system (although I understand there will be more
On Wednesday, November 20, 2013 3:56:51 PM UTC+1, Andrew Buchan wrote:
Further update:
Apache did finally crash, with the following message:
Server ran out of threads to serve requests. Consider raising the
ThreadsPerChild setting
I've bumped that up to 500, and there are less
Hi Niphlod, what is asking for disasters? Bumping it up to 500? Can you
elaborate on why? (I'm all for avoiding disasters given what's been
happening!)
The server-status page was showing 150 worker threads active, except lots
of the seem not to be from proper requests. Here's my post on
Hi Derek, the install on Apache is running fine and the ajax issue it threw
up I was able to fix as described above. So my problem is solved, but
thanks for checking in again.
I found that after updating to the latest version of web2py, it was easy to
get apache running following the deployment
That's cool, I was just asking about the side project of Niphlod's to look
at the pytds so that we could finally have a pure python database adapter.
With a pure python adapter, gevent could properly use greenlets to make
database queries, thus there would be no blocking on the queries.
On
any luck?
On Thursday, November 7, 2013 12:26:44 AM UTC-7, Niphlod wrote:
hard ? with DAL it's pretty easy if a module exposes the dbapi just
force the driver and implement the connect method and it's usually good to
go.
I'll test it when I get back home.
--
Resources:
-
I had a puzzling problem with 2.7.4 as a rocket service on Windows 2003,
but this was a problem with sqlform.grids. There was a print statement left
in code which caused the service to stop working. This cause doesn't match
with your insights into the problem, but I mention it anyway. You
Yes, I've been caught out by that quite a few times! Not a problem with
Apached I may add, the wsgi file redirects stdout to show up in the apache
error log, but I suppose you could point it anywhere really.
On Fri, Nov 8, 2013 at 12:31 PM, Tim Richardson t...@growthpath.com.auwrote:
I had a
Thanks for the suggestions, gevent looks good but I am indeed using pyodbc
and this seems to be working with Apache so I'd rather stick with this for
time being. (Note I had to embedd the Python27.dll manifest into pyodbc.pyd
to get it to work with Apache).
The reason for using ajax in the first
Check the logs, find the last line in the log - that's probably what caused
the freeze. Check your logging.conf to find out where the logs are kept.
On Tuesday, November 5, 2013 1:59:43 PM UTC-7, Andrew Buchan wrote:
Update:
I made a copy of the web2py installation on a new server (still
Hi Derek,
Thanks for replying, I had checked those but the last entry was months old,
so that was a dead end.
I finally got web2py running on Apache, though it took me till 5 am :-)
(ps: if anyone else plans on attempting to install Apache on Windows
against MSSQL, I'm happy to help out or write
Ah, if you are on Windows, I would recommend you run it with gevent instead
of rocket, it should be faster than even apache.
On Monday, November 4, 2013 4:15:56 AM UTC-7, Andrew Buchan wrote:
Hi,
I have a serious issue. Got a web2py install running as a service on a
Windows 2003 box
I'll also add this:
look at anyserver.py for how to get it going...
also, gevent does monkey.patch_all which means that all your tcp sockets
would become async, which means if you are freezing up because a single
connection is frozen, that won't be the case with gevent. Your individual
On Tue, Nov 5, 2013 at 11:53 AM, Andrew Buchan andyha...@gmail.com wrote:
Ok, the IT guy has disabled McAffee's on access scan for the folders
containing web2py stuff as well as python's installation directory. He
tells me that parts of McAffee other than on access scan may still
interfere
Yea, the ajax running synchronously - you provide callbacks because they do
run asynchronously.
To quote the jQuery book:
The* first letter in Ajax stands for asynchronous,* meaning that the
operation occurs in parallel and the order of completion is not guaranteed.
The async option to
On Thursday, November 7, 2013 12:23:06 AM UTC+1, Derek wrote:
Ah, if you are on Windows, I would recommend you run it with gevent
instead of rocket, it should be faster than even apache.
unfortunately it doesn't work as happily as it should: if you're using any
non-green module (he's
Yea, I'm working on a communication layer for pyodbc or pypyodbc which will
handle requests with gevent, thus your application should yield properly,
but it's not available just yet, and it won't necessarily be a drop-in
replacement for pyodbc though. In the meantime, this is usually good
It does look like there is a pure python implementation of TDS called
pytds, which may actually work... the hard part would be getting web2py to
use it...
On Wednesday, November 6, 2013 5:15:57 PM UTC-7, Derek wrote:
Yea, I'm working on a communication layer for pyodbc or pypyodbc which
hard ? with DAL it's pretty easy if a module exposes the dbapi just
force the driver and implement the connect method and it's usually good to
go.
I'll test it when I get back home.
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py
Ok, the IT guy has disabled McAffee's on access scan for the folders
containing web2py stuff as well as python's installation directory. He
tells me that parts of McAffee other than on access scan may still
interfere but there's not much we can do about that. This hasn't made any
difference.
Update:
I made a copy of the web2py installation on a new server (still pointing to
old database) and eventually got it set it up as a service but it still
freezes with no errors in web2py or in the event manager...
I'm pretty sure it's a programming error on my part somewhere or a
migration
Are you running Microsoft Endpoint Security? I have problems with the
virus scanner 'locking up' things under even light usage. One or two users
can bang all day, no problem but get more than 10 and it randomly freezes.
Our fix was to exclude pretty much anything Python related. YMMV.
On
Which version of web2py is it? Are you using gevent or rocket web server?
Can you try load testing your devel version and see if you can replicate
the issue? You can use something like siege or a similar tool.
Regards
On Monday, November 4, 2013 4:56:32 PM UTC+1, Andrew Buchan wrote:
Thanks
McAfee always seems to block things incorrectly, at least for me.
On Monday, November 4, 2013 8:56:32 AM UTC-7, Andrew Buchan wrote:
Thanks Willoughby,
We've got McAffee on that network, so have asked the IT guy to set it to
ignore python and web2py folders. Will let you know what the
It's running on rocket. I now think it's dying during a specific ajax call,
but only some of the time and/or shortly after its called (there's usually
a page load after it) and/or when it's excessively busy. Not sure if siege
can automate the button clicks that trigger the ajax calls but I
mdipierro wrote:
It should be completely backward compatible. Without the new version
is difficult for me to help debug the problem.
I would like to follow up on this, but I have been moved onto another
project for a while. It also seems that the replacement server is on
hold too, so I have
mdipierro wrote:
Do you see any issue with memory usage?
Overall it is difficult to tell as the server runs a few large
processes, including several big Java processes and a number of other
Python daemons, so overall memory usage (which is being graphed by
Cacti) is a bit all over the place
Candid wrote:
Every now and again there is a BUG: unable to handle kernel paging
request at virtual address b79b3000, always in the web2py Python
process, and web2py more or less hangs.
Where do you see this error message?
In /var/log/syslog.
I usually `grep Oops /var/log/syslog` to see
mdipierro wrote:
please upgrade to 1.79.2 and see if you still get the problem.
Massimo
This is a production system and downtime would be troublesome - are
there any differences between 1.76.5 and 1.79.2 that I should be aware
of? I have cron jobs, XML-RPC functions and a bunch of data entry
It should be completely backward compatible. Without the new version
is difficult for me to help debug the problem.
On 3 Lug, 02:34, Rowdy ro...@netspace.net.au wrote:
mdipierro wrote:
please upgrade to 1.79.2 and see if you still get the problem.
Massimo
This is a production system and
Candid wrote:
I am using web2py in my organization for internal applications. One of
the applications uses dedicated instance of web2py (v1.78.1) running
on dedicated ubuntu server. It connects to our SQL Server 2005
database via freetds. It runs with the following parameters:
python web2py.py
Do you see any issue with memory usage?
On 2 Lug, 03:52, Rowdy ro...@netspace.net.au wrote:
Candid wrote:
I am using web2py in my organization for internal applications. One of
the applications uses dedicated instance of web2py (v1.78.1) running
on dedicated ubuntu server. It connects to
Every now and again there is a BUG: unable to handle kernel paging
request at virtual address b79b3000, always in the web2py Python
process, and web2py more or less hangs.
Where do you see this error message?
On Jul 2, 4:52 am, Rowdy ro...@netspace.net.au wrote:
Candid wrote:
I am using
please upgrade to 1.79.2 and see if you still get the problem.
Massimo
On 2 Lug, 03:52, Rowdy ro...@netspace.net.au wrote:
Greetings,
I have a web2py application running in 1.76.5 (from source, and haven't
had the chance to test the app in a more recent version of web2py).
The server is
Can anyone shed light on this question:
Also, am I missing something or rocket should close the connections
if
request is taking longer than 10 seconds (default timeout)? So if I
put time.sleep(20) in my controller it should times out? That's not
what's happening - the controller waits all those
I will check memory usage next time it freezes but I am pretty sure
it's OK.
What do you mean by db log?
Also, am I missing something or rocket should close the connections if
request is taking longer than 10 seconds (default timeout)? So if I
put time.sleep(20) in my controller it should times
Here is Massimo's reply (sorry, I accidently hit Reply to author
button last time):
I am not familiar with SQLserver 2005 but probably it has a log file where to
store errors. If it running out of connections it
should say so there.
let me know.
No, the db server itself is fine. We have
Can you check memory usage when it freezes? Is there anything in the
db log about too many connections?
On 30 Giu, 16:06, Candid roman.bat...@gmail.com wrote:
I am using web2py in my organization for internal applications. One of
the applications uses dedicated instance of web2py (v1.78.1)
45 matches
Mail list logo
