The config I'm using is the one that comes from the web2py installer for
ubuntu/nginx/uwsgi in the scripts directory. So, if that were the case, I
think more people would be seeing it too.
-Jim
On Wed, Oct 26, 2016 at 9:06 AM, Richard Vézina wrote:
> Version of soft are not so differents so co
Version of soft are not so differents so config may still be used without
changes, but this should be double check... For instance, for me, config
had change significantly since 12.04 related to systemd... Nginx not much
but uwsgi has change. It may wise to check for deprecated uwsgi parameter
flag
mcm - thanks for that.
On Tuesday, October 25, 2016 at 12:23:27 PM UTC-5, mcm wrote:
>
> One clarification ... better performance is mostly due to the fact that
> the python GIL is not getting in the way
>
> 2016-10-25 19:20 GMT+02:00 Michele Comitini >:
>
>> > Regarding switching uwsgi to f
One clarification ... better performance is mostly due to the fact that the
python GIL is not getting in the way
2016-10-25 19:20 GMT+02:00 Michele Comitini :
> > Regarding switching uwsgi to fork-only - should I expect a performance
> hit?
>
> Unless you are very low on RAM you should expect
> Regarding switching uwsgi to fork-only - should I expect a performance
hit?
Unless you are very low on RAM you should expect better performance. Linux
fork is a clone call that uses fully fledged copy-on-write strategy. Uwsgi
forks are long running processes so forking overhead has little impa
Richard
Just checked my versions and my old systems that work have:
uwsgi 2.0.10
nginx 1.4.6
web2py 2.14.6
New system that needs altered uwsgi config
uwsgi 2.0.14
nginx 1.10.0
web2py 2.14.6
My uwsgi configs are identical except for the commented out 'limit-as =
512' line. Nginx configs are i
Jim which web server are we talking now? Nginx? I think you should
copy/paste configuration so we can have look, I can compare to mine... Just
the relevant part would suffice...
About threading notice of pandas it says holding lock in case of multiple
dataframe copy, but Jim report he can't even i
Massimo / Michele
Thanks for the input. I don't know anything about fork only mode. Looks
like I have some reading to do. Thanks again for the input.
-Jim
On Tue, Oct 25, 2016 at 4:02 AM, Michele Comitini <
michele.comit...@gmail.com> wrote:
> @Jim as per Massimo input, did you check that y
@Jim as per Massimo input, did you check that your uwsgi configuration is
in fork mode only, no threading? That could be one possible cause.
Try also gunicorn instead of uwsgi http://gunicorn.org/
2016-10-25 7:19 GMT+02:00 Massimo Di Pierro :
> Rocket is a multithreaded server and pandas is not
Rocket is a multithreaded server and pandas is not thread safe
(http://pandas.pydata.org/pandas-docs/stable/gotchas.html) be very careful
with it.
On Tuesday, 16 August 2016 13:19:42 UTC-5, Dave S wrote:
>
>
>
> On Tuesday, August 16, 2016 at 8:50:51 AM UTC-7, Jim S wrote:
>>
>> Hi
>>
>> I cur
Yes, it's crazy. I have 2 ubuntu 14.04 servers that work fine, pandas 0.18
(I have other issues with 0.19.0 so haven't moved there yet, but still had
the same behavior when I tested). But, any new Ubuntu 14.04 servers I
install won't work. Any time i try 16.10 or 16.04 (which is what i want to
u
You right, I forgot... But you may try by raising the limit... Is there any
open issue for pandas about that... This is not normal... If I remember it
does that only in web2py environment?
If the issue occurs on import pandas as pd why I don't have it??
I have : limit-as 512
I have pandas 0.19.0
Thanks for the comment Richard - At the point it crashes I don't believe it
is an issue with my app. It crashes on - import pandas as pd - so, I don't
think it is anything I'm doing in my code.
-Jim
On Mon, Oct 24, 2016 at 9:48 AM, Richard Vézina wrote:
> I suggest you try to raise the limit..
I suggest you try to raise the limit... I don't know why it get in the
configuration file, but I read it could be use to prevent memory leaks.
It's also help preventing app from eating all the memory before crash the
system... You maybe better understand why you code need so much memory and
try to
Started looking in to this issue again today and found this:
http://stackoverflow.com/questions/19439190/segmentation-fault-while-using-pandas-in-uwsgi
I took the advice of Roberto and removed the --limit-as in
/etc/uwsgi/web2py.ini - and now it seems to work find. I am clueless as to
what i
Thanks Dave
While my stated question was specific to haproxy, I'm really speaking in
more general terms as to whether or not the rocket server would be
sufficient behind ANY load balancer. I know I could have setup nginx as
the load balancer, but haproxy was really quite simple.
I'm going to rea
On Wednesday, August 17, 2016 at 1:15:58 PM UTC-7, Jim S wrote:
>
> Anyone else have thoughts on why I shouldn't use the rocket webserver
> behind haproxy?
>
>
Well, before today I didn't know anything about haproxy, but it seems to
have a good reputation as a load balancer. So it should take
Anyone else have thoughts on why I shouldn't use the rocket webserver
behind haproxy?
-Jim
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py (Source code)
- https://code.google.com/p/web2py/issues/list (Report Issues)
---
You received
Importing pandas from the web2py shell works just fine.
administrator@ubuntu16-2:/home/www-data/web2py$ sudo python web2py.py -a
password -S connect
web2py Web Framework
Created by Massimo Di Pierro, Copyright 2007-2016
Version 2.14.6-stable+timestamp.2016.05.10.00.21.47
Database drivers availabl
Strange!
Missing dependency which didn't install because of a broken link?
Richard
On Tue, Aug 16, 2016 at 3:22 PM, Jim Steil wrote:
> From my system:
>
> administrator@ubuntu16-2:/etc/nginx/sites-enabled$ sudo pip show pandas
> [sudo] password for administrator:
> ---
> Metadata-Version: 2.0
Which version of web2py?
I just try impor... with command line web2py instance and it works...
I am using web2py 2.14.6
Richard
On Tue, Aug 16, 2016 at 3:34 PM, Jim Steil wrote:
> Let me restate my issue
>
> I can use pandas just fine on ubuntu 16.04. The problem occurs when I'm
> importing
You may have a check, here some way of doing it :
http://stackoverflow.com/questions/16294819/how-to-check-if-my-python-has-all-required-packages
http://stackoverflow.com/questions/22213997/programmatically-check-if-python-dependencies-are-satisfied
Would try with pip freeze first...
Richard
O
Let me restate my issue
I can use pandas just fine on ubuntu 16.04. The problem occurs when I'm
importing a module into a web2py controller and that module fails to import
on the 'import pandas as pd' line.
Just to be clear, pandas works find on the box. It is when it runs on
web2py through the
>From my system:
administrator@ubuntu16-2:/etc/nginx/sites-enabled$ sudo pip show pandas
[sudo] password for administrator:
---
Metadata-Version: 2.0
Name: pandas
Version: 0.18.1
Summary: Powerful data structures for data analysis, time series,and
statistics
Home-page: http://pandas.pydata.org
Aut
Maybe try reinstall pip package, or if you use ubuntu package, use pip one
instead...
On Tue, Aug 16, 2016 at 3:16 PM, Richard Vézina wrote:
> Can you identify the responsible piece of pandas code that cause the
> issue? It don't make sens if you can't import pandas as pd in U16.04... I
> just t
Can you identify the responsible piece of pandas code that cause the issue?
It don't make sens if you can't import pandas as pd in U16.04... I just try
and it works...
sudo pip show pandas
---
Metadata-Version: 1.1
Name: pandas
Version: 0.18.1
Summary: Powerful data structures for data analysis,
Richard
I have not had it successfully running under ubuntu 16.04 with a prior
version of pandas. I have it running on Ubuntu 14.04 with the same version
of pandas. It fails on:
import pandas as pd
-Jim
On Tue, Aug 16, 2016 at 1:02 PM, Richard Vézina wrote:
> Massimo's has always said it wa
On Tuesday, August 16, 2016 at 8:50:51 AM UTC-7, Jim S wrote:
>
> Hi
>
> I currently have my app deployed using Ubuntu 14.04 / nginx / uwsgi. I
> have a couple of servers load-balanced behind a haproxy server. I'm
> running ssl on the haproxy system and talking http to web2y through
> nginx
Massimo's has always said it wasn't secure, I ignore the exact reason(s)...
I suggest downgrading pandas with this rational you were using is in
previous version of pandas and nginx without this problem...
I do use pandas (but not HDF5 Store) and I don't experiment this issue...
neither in 12.04
In my reading I didn't see that downgrading pandas would help. Am I missing
something or did I mislead you with my post? Downgrading ubuntu would help
(which is what we have now, but would like to run the latest ubuntu).
But, back to the main question, what are the biggest factors in not
recom
You can't downgrade pandas package?
Richard
On Tue, Aug 16, 2016 at 12:35 PM, Jim S wrote:
> Ron
>
> It would depend on what OS you're running on the x86 box. This would be a
> good place to start http://web2py.com/books/default/chapter/29/13/
> deployment-recipes
>
> Also, if you don't mind,
Ron
It would depend on what OS you're running on the x86 box. This would be a
good place to
start http://web2py.com/books/default/chapter/29/13/deployment-recipes
Also, if you don't mind, could you re-post this in a new thread so this
thread could focus on my initial question?
-Jim
On Tues
On that note, can someone point me to documentation about how to set up an
environment in dedicated sever on X86? I am thinking about co-locating than
shared hosting. If someone does go that route, how we manage to set up the
hosting environment?
On Tuesday, August 16, 2016 at 11:50:51 AM UT
33 matches
Mail list logo