On Fri, Feb 20, 2015 at 4:41 PM, Konstantin Kolinko <knst.koli...@gmail.com> wrote:
> 2015-02-21 0:10 GMT+03:00 Sean Dawson <seandawson2...@gmail.com>: > > We have a GWT app deployed to tomcat (7_59) and fairly often when we > send a > > bunch of request quickly we're seeing undefined methods in the logs - and > > the calls fail, causing issues with our app. We make calls via RestyGwt > > (latest version) but GwtRequests all show this - both though after a > number > > of REST calls in a short period of time. So for example... > > > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path1 HTTP/1.1" 200 > 304 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path2 HTTP/1.1" 200 > 310 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path3 HTTP/1.1" 200 > 307 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "undefinedDELETE /path4 > > HTTP/1.1" 501 304 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path5 HTTP/1.1" 200 > 304 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path6 HTTP/1.1" 200 > 310 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "DELETE /path7 HTTP/1.1" 200 > 307 > > [ip-addr] - - [20/Feb/2015:15:24:34 -0500] "undefinedDELETE /path8 > > HTTP/1.1" 501 304 > > > > Similarly... > > > > ... > > .... "undefinedPOST /gwtRequest HTTP/1.1" 501 1136 > > > > Very little info online, but did come across this old bug... > > > > https://bz.apache.org/bugzilla/show_bug.cgi?id=49779 > > > > In fiddler, the headers are identical between the requests that work and > > those that fail. Resending the failed request completes normally. > > > > So far we've only be able to reproduce this when using Internet Explorer > > (10 & 11) and we've spent a lot of time trying to figure out what's going > > on - but have been unable. Any pointers/explanations? > > > > Thanks! > > "undefined" is a JavaScript word. In Java I would expect "null" > instead of that word. > > > In fiddler, the headers are identical between the requests that work and > > those that fail. > > The string in access log is not a header. It is HTTP request line. > The first line of an HTTP request. > > Ok, but this is in the standard tomcat access logs, using standard logging, and is in the method name, not URL. Maybe I'm not understanding what you're saying here. > BTW, a similar issue at stackoverflow (but the "undefined" string was > added to URL part of request line): > > > http://stackoverflow.com/questions/11017609/undefined-randomly-appended-in-1-of-requested-urls-on-my-website-since-12-jun > Title: “undefined” randomly appended in 1% of requested urls on my > website since 12 june 2012 > > We did come across it but again our's is in the method, not in the URL. > > One of theories there is that some browser addon was malfunctioning. > > Ok, this has happened on about 5 people's machines with a couple different versions of IE - I don't think we have any addons at all in some cases. > If nothing else helps, it should be easy to implement a Valve for > Tomcat that will fix the wrong request.getMethod() value before > passing it to a web application. > > I don't know much about that but we could give it a try - so.... someone else is changing the method somewhere before it gets to tomcat? and the Valve will change it back? Thanks. Best regards, > Konstantin Kolinko > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > For additional commands, e-mail: users-h...@tomcat.apache.org > >