Really? Hmm.
<Location /examples>
ProxyPass http://localhost:8080/examples
ProxyPassReverse http://localhost:8080/examples
ProxyPass ws://localhost:8080/examples
ProxyPassReverse ws://localhost:8080/examples
</Location>
Debug did show something interesting. This was 2 browsers, Firefox and Chrome
at the same time.:
[Mon Mar 18 19:32:47.311154 2013] [proxy_wstunnel:debug] [pid 16104:tid
4487745536] mod_proxy_wstunnel.c(228): [client ::1:61267] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:47.311197 2013] [proxy_wstunnel:debug] [pid 16104:tid
4487745536] mod_proxy_wstunnel.c(237): [client ::1:61267] AH02446: sock was
readable
[Mon Mar 18 19:32:47.311415 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(228): [client ::1:61271] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:47.311434 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(237): [client ::1:61271] AH02446: sock was
readable
[Mon Mar 18 19:32:47.397291 2013] [proxy_wstunnel:debug] [pid 16104:tid
4487745536] mod_proxy_wstunnel.c(228): [client ::1:61267] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:47.397341 2013] [proxy_wstunnel:debug] [pid 16104:tid
4487745536] mod_proxy_wstunnel.c(253): [client ::1:61267] AH02448: client was
readable
[Mon Mar 18 19:32:47.397352 2013] [reqtimeout:info] [pid 16104:tid 4487745536]
[client ::1:61267] AH01382: Request body read timeout
[Mon Mar 18 19:32:47.397359 2013] [proxy_wstunnel:debug] [pid 16104:tid
4487745536] mod_proxy_wstunnel.c(129): (70007)The timeout specified has
expired: [client ::1:61267] AH02442: error on client - a
p_get_brigade
[Mon Mar 18 19:32:47.397379 2013] [proxy:debug] [pid 16104:tid 4487745536]
proxy_util.c(2028): AH00943: WS: has released connection for (localhost)
[Mon Mar 18 19:32:47.410563 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(228): [client ::1:61271] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:47.410618 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(237): [client ::1:61271] AH02446: sock was
readable
[Mon Mar 18 19:32:47.511173 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(228): [client ::1:61271] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:47.511220 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(237): [client ::1:61271] AH02446: sock was
readable
[
[Mon Mar 18 19:32:51.180493 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(228): [client ::1:61271] AH02445: woke from
poll(), i=1
[Mon Mar 18 19:32:51.180545 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(253): [client ::1:61271] AH02448: client was
readable
[Mon Mar 18 19:32:51.180558 2013] [reqtimeout:info] [pid 16105:tid 4487208960]
[client ::1:61271] AH01382: Request body read timeout
[Mon Mar 18 19:32:51.180565 2013] [proxy_wstunnel:debug] [pid 16105:tid
4487208960] mod_proxy_wstunnel.c(129): (70007)The timeout specified has
expired: [client ::1:61271] AH02442: error on client - ap_get_brigade
[Mon Mar 18 19:32:51.180585 2013] [proxy:debug] [pid 16105:tid 4487208960]
proxy_util.c(2028): AH00943: WS: has released connection for (localhost)
100% Reproducible, and when accessing Tomcat7 directly (no httpd), the sockets
are verified okay for 7hrs. (left running while away).
Any insight? mod_reqtimeout?
On 2013-03-18, at 1:50 PM, Jim Jagielski wrote:
> I cannot recreate either. Anything in the log (debug level) which
> may be pertinent?
>
> On Mar 18, 2013, at 12:43 PM, Jamie Johnson <[email protected]> wrote:
>
>> I haven't done any serious testing but haven't seen this in my simple
>> application. I will try to duplicate the issue with the snakes example and
>> see where I land.
>>
>> Perhaps this is something that one of the devs can shed more light on though.
>>
>>
>> On Mon, Mar 18, 2013 at 11:59 AM, Nathan Quinlan <[email protected]>
>> wrote:
>> Seems to work fine with the correct Location settings. ;)
>>
>> However I have found that when proxying to Tomcat 7 (what I'm testing with)
>> the tunnel lasts for about 20 seconds or so and then terminates.
>> After a few reloads it appears as though Tomcat still has the socket open
>> (as timeout I setup to be 5 minutes) but the connection appears to be
>> failing in httpd.
>>
>> This connection shutdown appears to occur within httpd because when I
>> apachectl -k restart the browsers connection resumes as expected.
>> In this case I'm testing Tomcat's /examples/websocket/snakes.html and the
>> player state hangs in < 30 seconds and browser thinks socket is closed BUT
>> when restarting httpd the connection resumes as far as Tomcat is concerned.
>>
>> I've taken a capture of the traffic using Wireshark for browser <-> httpd
>> and there isn't anything odd in the last packet sent, it just stops.
>>
>> Any ideas? The approximate 20 second cut off seems odd and httpd is
>> basically stock 2.4.4 compiled with the merged in wstunnel module.
>>
>> On 2013-03-18, at 7:29 AM, Jim Jagielski wrote:
>>
>>> ProxyPass /whatever ws://websocket-srvr.example/com/
>>>
>>> Basically, the new submodule adds the 'ws' and 'wss' scheme
>>> to the allowed protocols between the client and the backend, so
>>> you tell Apache that you'll be talking 'ws' with the
>>> backend (same as ajp://whatever sez that httpd will be
>>> talking ajp to the backend).
>>>
>>> On Mar 17, 2013, at 9:52 AM, Jamie Johnson <[email protected]> wrote:
>>>
>>>> I am able to duplicate your issue, not 100% sure what's happening there or
>>>> if that configuration is supposed to be supported or not. The following
>>>> works fine though
>>>>
>>>> <Location /dynamic>
>>>> ProxyPass http://10.0.1.11:8080/WebSockets
>>>> ProxyPassReverse http://10.0.1.11:8080/WebSockets
>>>> </Location>
>>>> <Location /dynamic/ws>
>>>> ProxyPass ws://10.0.1.11:8080/WebSockets/ws
>>>> ProxyPassReverse ws://10.0.1.11:8080/WebSockets/ws
>>>> </Location>
>>>>
>>>> would be good to hear exactly how this is expected to be configured,
>>>> perhaps we're landing in the wrong portion of the code when they are in
>>>> the same location?
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Sun, Mar 17, 2013 at 9:43 AM, Jamie Johnson <[email protected]> wrote:
>>>> So your setup is a little different than mine, I have the following and
>>>> even with images it works
>>>>
>>>> ProxyPass /ws ws://10.0.1.11:8080/WebSockets/ws
>>>>
>>>> ProxyPassReverse /ws ws://10.0.1.11:8080/WebSockets/ws
>>>>
>>>>
>>>> ProxyPass /test http://10.0.1.11:8080/WebSockets
>>>>
>>>> ProxyPassReverse /test http://10.0.1.11:8080/WebSockets
>>>>
>>>> I will try with a similar setup to yours now and see where I get.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Sun, Mar 17, 2013 at 9:19 AM, Jamie Johnson <[email protected]> wrote:
>>>> Hmmm... I am only serving one html file in my test...I will add some more
>>>> and see if I notice the same
>>>>
>>>> On Mar 17, 2013 8:57 AM, "Nathan Quinlan" <[email protected]> wrote:
>>>> I don't have any rewrite rules.
>>>> <Location /dynamic>
>>>> #ProxyPass ajp://localhost:8009/dynamic
>>>> ProxyPass http://localhost:8080/dynamic
>>>> ProxyPass ws://localhost:8080/dynamic
>>>>
>>>> And
>>>> Alias "/static" "/opt/sitestatic"
>>>> <Directory "/opt/sitestatic">
>>>>
>>>> Now, with the ProxyPassReverse present about 50% of the files transfered
>>>> get that 404 -as- served by Tomcat (based on Apache-Coyote/1.1 header).
>>>> Despite the URL being something like /static/hello.jpg.
>>>>
>>>> On 2013-03-17, at 6:33 AM, Jamie Johnson wrote:
>>>>
>>>>> Also I have a different rewrite rule for http and ws...not sure that is
>>>>> required though
>>>>>
>>>>> On Mar 17, 2013 6:31 AM, "Jamie Johnson" <[email protected]> wrote:
>>>>> Definitely interesting. I had gotten this working but I was only serving
>>>>> sine html through tomcat and I made the endpoints different while testing
>>>>> even though they were coming from the same application in tomcat...mine
>>>>> are birth being proxied as well no rewrite involved
>>>>>
>>>>> On Mar 17, 2013 12:35 AM, "Nathan Quinlan" <[email protected]>
>>>>> wrote:
>>>>> I do not know why but for some reason ProxyPassReverse doesn't play nice
>>>>> with Tomcat and I had to include an extra line for the ws: protocol.
>>>>>
>>>>> Additionally with the ProxyPassReverse present I would see crazy response
>>>>> 404 headers when loading say 30 small images on screen like:
>>>>> Content-Length 1003
>>>>> Content-Type text/html;charset=utf-8
>>>>> Date Sun, 17 Mar 2013 04:09:28 GMT
>>>>> Server Apache-Coyote/1.1
>>>>>
>>>>> The interesting thing was that in this case the image being loaded had a
>>>>> totally different URL (/a proxy to Tomcat, /b static content) and was
>>>>> handled outside of Tomcat via a rewrite rule and a <Location> but when
>>>>> the ProxyPassReverse was removed images were fine.
>>>>> Images that were not 404 show up in the access_log of httpd but the 404
>>>>> files with the crazy header do not.
>>>>>
>>>>>
>>>>>
>>>>> On 2013-03-16, at 12:37 PM, Jamie Johnson wrote:
>>>>>
>>>>>> I just took a quick stab and it was pretty straight forward, I just
>>>>>> added lines like this and it appeared to work properly
>>>>>>
>>>>>> ProxyPass /ws http://hostname:port/ws/websocket
>>>>>> ProxyPassReverse /ws http://hostname:port/ws/websocket
>>>>>>
>>>>>> again, this appeared to work properly I am next going to be giving SSL a
>>>>>> try to see if things work properly with that. If what I did above is
>>>>>> not right any info would be appreciated. Also should I expect the SSL
>>>>>> support to work?
>>>>>>
>>>>>>
>>>>>> On Sat, Mar 16, 2013 at 12:21 PM, Jamie Johnson <[email protected]>
>>>>>> wrote:
>>>>>> I've just built the latest code on trunk to test proxy_wstunnel, but
>>>>>> haven't seen any documentation on how to configure it. Is this
>>>>>> available anywhere?
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>
>>
>
