I was trying to write myself a step-by-step procedure for myself to create a local development copy of a website. Basically, I was using Git to create a repo on the production site and push that to github. Then clone that repo to my local computer as dev environment, rsync the sites/default/files and import a sql-dump into the local DB.
Well, it would not work. I could go to standard web pages, but if I tried to go to admin pages (like admin/reports/status), I would get an error in Chrome Like this:
Error 101 (net::ERR_CONNECTION_RESET): The connection was reset.
Windows event log reported:
Faulting application name: httpd.exe, version: 2.2.22.0, time stamp: 0x4faff1bad
Faulting module name: php5ts.dll, version: 5.3.12.0, time stamp: 0x4fb027d7
Exception code: 0xc00000fd
Oddly, if I just downloaded the files via SCP to my local machine (and did all the other stuff) the entire local website would work as expected.
I spent hours determining that it was ONE FILE from the Print 7.x-1.x module (print_pdf.install) that was causing the problem. I could clone the repo from github, but all the admin pages would cause the error. If I just copied that one file via scp, the local website would work.
I compared the two versions of the file in notepad++ and they were reportedly identical.
I tried to do a git diff on the file after coping over the "good" file, but only got some output about CRLF being converted to LF but my working directory was okay. I don;t recall the exact output.
I spent hours trying to figure it out and finally noticed that there was a new version of the Print module, so I upgraded, pushed to github, pulled the new version and it worked without needing to scp the print_pdf.install file.
So basically, a single file, having gone into and back out of a git repo, became different enough from the original to crash the local website.
In case I run into this again, does anyone know what might have been happening?