ikiwiki/doc/bugs/Error:_OpenID_failure:_time_bad_sig:.mdwn

Sometimes when I try to login (to edit a page) here then I can see
the following error message:

    Error: OpenID failure: time_bad_sig:

Please note that authorization process successes on OpenID server side.

It occurs both for [getopenid.com](http://www.getopenid.com/)
and [myopenid.com](http://www.myopenid.com/) servers I use.

I'm reporting this, but I'm not sure whether a problem is with your
ikiwiki or my OpenID servers. --[[Paweł|ptecza]]

> I've seen this too, once or twice (using myopenid), and reauthenticating
> fixed it -- so I can't reproduce it reliably to work on it. I think I've
> seen it both on this wiki and on the one running on my laptop.
> 
> The perl openid client module seems
> to fail with time_bad_sig if the time in the signature from the other end
> is "faked". I'm not 100% sure what this code does yet:

	# check age/signature of return_to
	my $now = time();
	{
		my ($sig_time, $sig) = split(/\-/, $self->args("oic.time") || "");
		# complain if more than an hour since we sent them off
		return $self->_fail("time_expired")   if $sig_time < $now - 3600;
		 also complain if the signature is from the future by more than 30 seconds,
		# which compensates for potential clock drift between nodes in a web farm.
		return $self->_fail("time_in_future") if $sig_time - 30 > $now;
		# and check that the time isn't faked
		my $c_secret = $self->_get_consumer_secret($sig_time);
		my $good_sig = substr(OpenID::util::hmac_sha1_hex($sig_time, $c_secret), 0, 20);
		return $self->_fail("time_bad_sig") unless $sig eq $good_sig;
	}

> At least it doesn't seem to be a time sync problem since the test for too
> early/too late times have different error messages.. --[[Joey]]

I've had this problem too, but with my track record of reporting OpenID bugs 
I thought it best if I held my tongue. I usually experience this the first
time I sign in on any ikiwiki installation of {ikiwiki.kitenet, ikidev, 
betacantrips}, and I think re-logging in always works. --Ethan

> Does seem easier to repro than I thought.
> Ok, fixed it.. done --[[Joey]] (reopened for new instance of same error
> message below)

----

## the return of the nasty bug

Hmmmmm, looks like it is not entirely fixed.  I am getting it on my own
[blog](http://blog.tobez.org/).  Just upgraded to 3.20110430, same same.
I am using custom openid with redirection to myopenid.com.
Please tell me if you need more info.  The same openid worked fine to login to *this* site to post this.
-- Anton

> Well, this bug is from 2007. Probably you are not encountering the same
> bug.
> 
> I also have a openid delegation to myopenid, and I can reproduce the
> problem when logging into your site.
>
> What version of the
> Net::OpenId::Consumer perl library do you have installed? --[[Joey]] 

>> It is the latest version from FreeBSD's ports collection,
>> which happens to be a [slightly patched up](http://www.freebsd.org/cgi/cvsweb.cgi/~checkout~/ports/net/p5-Net-OpenID-Consumer/files/patch-Consumer.pm?rev=1.2;content-type=text%2Fplain)
>> variant of an
>> ["unauthorized" CPAN release 1.06](http://search.cpan.org/~gugu/Net-OpenID-Consumer-1.06/)
>>
>> Do you think it might be a good idea to try with 1.03 or with an unpatched 1.06?
>> -- Anton

>>> Absolutely. --[[Joey]] 

>>>> 1.03 fails with "Error: login failed, perhaps you need to turn on cookies?" (needless to say cookies are enabled).  
>>>> Unpatched 1.06 fails with "Error: login failed, perhaps you need to turn on cookies?".  
>>>> 1.03 with the same patch fails with "Error: OpenID failure: time_bad_sig:" -- Anton.

>>>>> Investigation revealed it was a bug in the freebsd patch, which I
>>>>> understand is going to be dealt with. [[done]] --[[Joey]] 

I am getting the same error here with ikiwiki 3.20120629 (wheezy). I had trouble with ikiwiki-hosting configurations of OpenID, basically related to the `openid_realm` parameter - which I had to comment out. But now it seems to fail regardless. --[[anarcat]]

> Nevermind, this was because I was blocking cookie on the CGI (!!). Message *could* be improved though, it's not the first time i stumble upon this... --[[anarcat]]