
In article
In article <3D8D97FDA7A52F4CB773EF3BBA9470D3106924@svgw2k15.roxardomain.roxar.com>, "Thomas Nilsen"
wrote: You're quite right. The httpd daemon should not crash as a result - but in any case, it could give rather strange results. In the cases i've seen and tested, IE would return a DNS error - which if course had nothing to do with it. But anyway, I suggest you bump up the error logging on Apache and see if that can give you any clues.
thomas
Initial testing: I reduced the $Self->{WebMaxFileUpload} setting in OTRS from 5Mb to 4Mb. Now the http daemon thrashes and is killed when I try to attach the 4.8Mb document.
This variable seems to be only used to set the value of $CGI::POST_MAX in Kernel/System/Web/Request.pm - time for some more Googling!
Googling for $CGI::POST_MAX brings up a number of similar occurrences which are nothing to to with OTRS. This URL http://www.codecomments.com/message162736.html details the problem, which occurs in CGI.pm version 3.0 and higher - it also has a diff of the allegedly offending code. I tried backing out the code in CGI.pm and the apache daemon no longer consumes all the memory and gets killed, but the user just eventually gets a "Couldn't connect to server" error rather than a sane error message about file size restrictions. Further poking in OTRS reveals that although an error routine exists in Request.pm, I can't seem to find where it is called! Dave -- Dave Hill Newnham Research Ltd