I have a new centos installation. I followed the script at
http://web2pyslices.com/main/slices/take_slice/98. When I use the new
application wizard, I get the same traceback that others have
posted:
Traceback (most recent call last):
File "/opt/web-apps/web2py/gluon/restricted.py", line 188, in
restricted
exec ccode in environment
File "/opt/web-apps/web2py/applications/admin/controllers/
wizard.py", line 572, in <module>
File "/opt/web-apps/web2py/gluon/globals.py", line 95, in <lambda>
self._caller = lambda f: f()
File "/opt/web-apps/web2py/applications/admin/controllers/
wizard.py", line 200, in step6
create(form.vars)
File "/opt/web-apps/web2py/applications/admin/controllers/
wizard.py", line 489, in create
file=open(meta,'wb')
IOError: [Errno 2] No such file or directory: '/opt/web-apps/web2py/
applications/admin/../xxx/wizard.metadata'
When I use the "New Simple Application" part of the admin page, all I
get is a "Flash" that says, "unable to create application <appname>"
I've poked around in the code, and I believe that the problem takes
place a few lines BEFORE the offending line 489, as follows:
On line 485, we find
"app_create(app,request,force=True,key=params['security_key'])"
App_create, which is defined in gluon/admin.py, needs to create the
directory in applications to hold the app. This function returns True
or False, but line 485 does not check the return value.
My first thought was that the os.mkpath in app_create was failing. I
still believe this, since the directory is not being created.
I checked the permissions. The applications directory is owned by
'apache'. The web server is running as apache. Even WSGI is set to
run as apache. So how can os.mkdir(path) fail?
No errors are generated in the web server's error log for this, and
the only clue I have is the traceback when using the wizard.
Any help appreciated.
