invirt-reload fails on xvm-dev
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
XVM |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
root@xvm-dev:~# invirt-reload
* Reloading config for invirt-
* Reloading PostgreSQL 8.3 database server [ OK ]
* Restarting the Invirt DNS server invirt-dns [ OK ]
* Reloading Invirt VNC client certificates invirt-vnc-client Certificate was added to keystore
Invalid hostname specified
keytool error: java.lang.
* Reloading config for invirt-web Traceback (most recent call last):
File "/usr/bin/
main()
File "/usr/bin/
print render(data)
File "/usr/bin/
return Template(data, lookup=
File "/usr/lib/
return runtime.
File "/usr/lib/
_render_
File "/usr/lib/
_exec_
File "/usr/lib/
callable_
File "memory:
File "/var/lib/
% (prefix, key))
invirt.
* Stopping web server apache2 [ OK ]
Stopping kstart for Apache2: k5start.
Starting kstart for Apache2: k5startNote: Operation is performed on cell sipb.mit.edu
.
* Starting web server apache2 [ OK ]
* Reloading Invirt web server iptables rules invirt-web-iptables [ OK ]
On Fri, Feb 06, 2009 at 10:09:51AM -0000, Anders Kaseorg wrote: mako-render" , line 38, in <module> common. InvirtConfigErr or: missing configuration variable svn
> * Reloading config for invirt-web Traceback (most recent call last):
> File "/usr/bin/
> [...]
> invirt.
> [fail]
This is known and harmless, though we should fix it to stop throwing
an error. There's no svn stanza in xvm-devconfig, because it doesn't
serve a SVN repo, and there shouldn't particularly be. The
consequence of the error is that sites-available/svn is empty, which
has exactly the right effect.
The ultimate right thing here is that we should have a way of apache/ sites.d/ or something be copied into
inserting random stuff into the Apache config other than by putting it
in Mako in invirt-web and giving some pieces of it names in the
config. For the svn case this is fairly straightforward; let
/etc/invirt/
sites-available.
The other random hack we have in our web config is the Trac redirect;
supporting that would be a bit more complex. But since we don't use
Trac -- here we are -- we should just cut it.
Greg