On Tue, Jan 4, 2011 at 5:41 PM, Fabio Mascarenhas <mascaren...@acm.org>wrote:
> If you want to run just a single, persistent WSAPI application it is much
> easier to use this as your idlocutor.fcgi script:
>
> ----
> pcall(require,"luarocks.require")
>
> local fastcgi = require "wsapi.fastcgi"
> local myapp = require "myapp"
>
> fastcgi.run(myapp.run)
> ----
>
> Where myapp.run is your WSAPI entry point (that takes a WSAPI environment).
>
That's a nice idea, thanks Fábio.
>
> BTW, what version of WSAPI are you using? The common loader had a bug.
>
I'm using the version 1.3.4.
I'm going to try your idea and see if any other problem arises.
Best regards,
Katcipis
>
> --
> Fabio Mascarenhas
>
>
> On Tue, Jan 4, 2011 at 4:32 PM, Tiago Katcipis <katci...@inf.ufsc.br>wrote:
>
>> On Mon, Jan 3, 2011 at 11:27 AM, Tiago Katcipis <katci...@inf.ufsc.br>wrote:
>>
>>>
>>>
>>> On Mon, Jan 3, 2011 at 11:15 AM, Fabio Mascarenhas
>>> <mascaren...@acm.org>wrote:
>>>
>>>> With isolated=true WSAPI will run your scripts in a pool of Rings, and
>>>> will collect and respawn the Rings periodically,
>>>>
>>>
>>> besides periodically, on any 4XX and on DELETE method it will be
>>> respawned too? just for curiosity :-).
>>>
>>>
>>>> so isolated=false is the behavior you want, as you are only running a
>>>> single script anyway.
>>>>
>>>
>>> everything is working ok now, it took some time and a good read on the
>>> source to understand this clearly.
>>>
>>
>> With a little further testing...things seems to not be that ok after all.
>>
>> After making a lot of requests (GET, POST, DELETE) wsapi starts to respond
>> the following:
>>
>> *<html>
>> <head><title>WSAPI Error in Application</title></head>
>> <body>
>> <p>There was an error in the specified application.
>> The full error message follows:</p>
>> <pre>
>> /usr/lib/lua/5.1/wsapi/common.lua:78: not a valid WSAPI application
>> stack traceback:
>> [C]: in function 'error'
>> /usr/lib/lua/5.1/wsapi/common.lua:78: in function
>> </usr/lib/lua/5.1/wsapi/common.lua:65>
>> (tail call): ?
>> (tail call): ?
>> /usr/lib/lua/5.1/wsapi/common.lua:710: in function
>> 'load_wsapi_persistent'
>> /usr/lib/lua/5.1/wsapi/common.lua:728: in function
>> </usr/lib/lua/5.1/wsapi/common.lua:721>
>> (tail call): ?
>> (tail call): ?
>> [C]: in function 'xpcall'
>> /usr/lib/lua/5.1/wsapi/common.lua:254: in function 'run_app'
>> /usr/lib/lua/5.1/wsapi/common.lua:281: in function 'run'
>> /usr/lib/lua/5.1/wsapi/fastcgi.lua:40: in function 'run'
>> ./idlocutor.fcgi:20: in main chunk
>> [C]: ?
>> </pre>
>> </body>
>> </html>
>>
>> *It happens randomly (taking a look at the debug of my fastcgi, no error
>> is reported, everything is working fine, suddenly this error starts and my
>> fastcgi is not even called), and what could mean "not a valid WSAPI
>> application" ??? My lua script stopped to be a WSAPI application ?
>>
>> It does not try to reload the script, the only way i can get out of it is
>> by killing the fastcgi process (wsapi.fastcgi). So the server starts a new
>> one and everything goes back to normal...until the problem happens again.
>> *
>>
>> *someone has a hint of what this could be? it started to happens with the
>> isolated = false option.*
>>
>> *Best regards,
>> Katcipis*
>>
>> *
>>
>>>
>>> Thanks for the help,
>>> Katcipis
>>>
>>>
>>>>
>>>> --
>>>> Fabio Mascarenhas
>>>>
>>>> On Wed, Dec 29, 2010 at 2:48 PM, Tiago Katcipis
>>>> <katci...@inf.ufsc.br>wrote:
>>>>
>>>>>
>>>>>
>>>>> On Wed, Dec 29, 2010 at 11:31 AM, Tiago Katcipis <katci...@inf.ufsc.br
>>>>> > wrote:
>>>>>
>>>>>> Hi all,
>>>>>>
>>>>>> I'm having some problem using wsapi-fastcgi with lighthttpd. I got
>>>>>> everything working fine, but every time an error occurs (any 4xx) or
>>>>>> someone
>>>>>> send a DELETE method request my script is reloaded. I don't want the
>>>>>> script
>>>>>> to reload because every time this happens i reconnect to the db and the
>>>>>> older connections get lost and the db continues to consume resources (on
>>>>>> this case is mongo db, a thread gets locked forever each time my script
>>>>>> is
>>>>>> reloaded).
>>>>>>
>>>>>> When there is no error 4XX or DELETE everything works as expected, the
>>>>>> script is not reloaded on each request and the connection is reused (i
>>>>>> run
>>>>>> an init function when the script is loaded, but it is supposed to run
>>>>>> only
>>>>>> once and them handle the requests without reloading).
>>>>>>
>>>>>> Here is the configuration that i run the fastcgi:
>>>>>>
>>>>>> #!/usr/bin/lua
>>>>>> local common = require "wsapi.common"
>>>>>> local fastcgi = require "wsapi.fastcgi"
>>>>>>
>>>>>> local ONE_HOUR = 60 * 60
>>>>>> local ONE_DAY = 24 * ONE_HOUR
>>>>>>
>>>>>> local wsapi_loader = common.make_loader{
>>>>>> isolated = true, -- isolate each script in its own Lua
>>>>>> state
>>>>>>
>>>>>
>>>>> Taking a good look at the code i just changed the isolated = true to
>>>>> false, so it would call the make_persistent_loader instead of the
>>>>> isolated.
>>>>> It worked fine just as i expected, still don't know why with isolated =
>>>>> true
>>>>> it reloads the script when returning a 4XX or receiving a DELETE.
>>>>>
>>>>> I'm working with only one script, so the problem seems resolved to me.
>>>>>
>>>>> Sorry for the noise.
>>>>>
>>>>> Best regards,
>>>>> Katcipis
>>>>>
>>>>>
>>>>>> filename = nil, -- if you want to force the launch of a single
>>>>>> script
>>>>>> launcher = "idlocutor.fcgi", -- the name of this script
>>>>>> reload = false, -- if you want to reload the
>>>>>> application on every request
>>>>>> period = ONE_HOUR, -- frequency of Lua state staleness
>>>>>> checks
>>>>>> ttl = ONE_DAY, -- time-to-live for Lua states
>>>>>> vars = -- order of checking for the path of
>>>>>> the script
>>>>>> { "SCRIPT_FILENAME",
>>>>>> "PATH_TRANSLATED" }
>>>>>> }
>>>>>>
>>>>>> fastcgi.run(wsapi_loader)
>>>>>>
>>>>>>
>>>>>> i tried to take a look on wsapi.fastcgi and them on wsapi.commom to
>>>>>> understand why it could be reloading my script, i understood that when
>>>>>> the
>>>>>> file is modified or reload is true it will reload the script on a new
>>>>>> lua_State, but i was unable to find another moment when this is
>>>>>> done..related to the errors (the rings logic is not too much easy to
>>>>>> understand, specially to me, never used rings).
>>>>>>
>>>>>> this reload problem (or maybe expected behaviour....i don't know) is
>>>>>> caused by wsapi? wsapi-fastcgi? or is something related to lighthttpd? (i
>>>>>> doubt of that because the pid of the fastcgi process remains the
>>>>>> same...it
>>>>>> seems that wsapi is reloading the script internally) I'm kinda new to
>>>>>> web
>>>>>> development, sorry I'm asking something stupid.
>>>>>>
>>>>>>
>>>>>> best regards,
>>>>>> Katcipis
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> http://www.getgnulinux.org/windows
>>>>>
>>>>> _______________________________________________
>>>>> Kepler-Project mailing list
>>>>> Kepler-Project@lists.luaforge.net
>>>>> http://lists.luaforge.net/cgi-bin/mailman/listinfo/kepler-project
>>>>> http://www.keplerproject.org/
>>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Kepler-Project mailing list
>>>> Kepler-Project@lists.luaforge.net
>>>> http://lists.luaforge.net/cgi-bin/mailman/listinfo/kepler-project
>>>> http://www.keplerproject.org/
>>>>
>>>
>>>
>>>
>>> --
>>> http://www.getgnulinux.org/windows
>>>
>>
>>
>>
>> --
>> http://www.getgnulinux.org/windows
>>
>> _______________________________________________
>> Kepler-Project mailing list
>> Kepler-Project@lists.luaforge.net
>> http://lists.luaforge.net/cgi-bin/mailman/listinfo/kepler-project
>> http://www.keplerproject.org/
>>
>
>
> _______________________________________________
> Kepler-Project mailing list
> Kepler-Project@lists.luaforge.net
> http://lists.luaforge.net/cgi-bin/mailman/listinfo/kepler-project
> http://www.keplerproject.org/
>
--
http://www.getgnulinux.org/windows
_______________________________________________
Kepler-Project mailing list
Kepler-Project@lists.luaforge.net
http://lists.luaforge.net/cgi-bin/mailman/listinfo/kepler-project
http://www.keplerproject.org/
