The token supplied to the function is invalid #107
Comments
|
Affected test cases... waffle-tests waffle-tomcat5/6/7/8 All 6 of these are using "Negotiate" for security package. If I switch these all to "NTLM", everything works. Please advice if this is something that should be changed or if this potentially indicates some other issue. Quick test of project jar files in tomcat worked without incident so I'm thinking test case issue only. But unclear as to what this started occurring when it didn't seem to before. Possibly machine related. |
|
Make a PR where these tests aren't ignored anymore. I'll take a look. |
|
I am not seeing failures in these tests locally (Windows 7 VM). However it is likely a real problem. Is your machine joined to a domain? What name(s) are being passed into things like |
|
Although all builds work even for me, I have the unit tests ignoring this error essentially. The stack trace is sent out to the log. This doesn't seem to have any impact to running the project components. com.sun.jna.platform.win32.Win32Exception: The token supplied to the function is invalid |
|
Tried full build out on another machine. This issue is not present there so likely a machine specific issue. Since I've already coded around the problem by accepting this condition, closing out this issue as it does not hinder any build from working now. |
A fix to capture tomcat 6, 7, and 8 negotiate authenticator Win32Exception which started occurring around June 2015 per issue #107. Around that time Exception was switched out for IOException which allowed Win32Exception to occur. Now catch that and perform same actions as IOException. Left original unit tests capturing fall through and added new check as this shows Negotiate, NTLM where the comma was normally expected to be a space. This does fix the condition but I'm not 100% certain this is correct from unit test side.
A fix to capture tomcat 6, 7, and 8 negotiate authenticator Win32Exception which started occurring around June 2015 per issue #107. Around that time Exception was switched out for IOException which allowed Win32Exception to occur. Now catch that and perform same actions as IOException. Left original unit tests capturing fall through and added new check as this shows Negotiate, NTLM where the comma was normally expected to be a space. This does fix the condition but I'm not 100% certain this is correct from unit test side.
|
I just bumped into this. The original issue seems to occur when client is on different domain than the web server is. When client is on same domain as server, everything works ok. Easy test case is to create a local username to workstation and try to access web server in domain. And although the this fix takes the stackdump away, it is not possible to login. I think that this worked OK in waffle 1.6. |
|
If the client is in a different domain and the domains don't have trust it can't work, because the server doesn't know anything about this user. I don't see how it worked in Waffle 1.6. |
|
What happens is that IE asks for username/password and if you enter But to summarize: But the last test was done using Win 10 preview desktop, I'll have to try with W7 or 8. I'll report back when I know more. |
|
Tested again with 1.7.4-SNAPSHOT (compiled today). Server is Win2012 on domain "A". Client is Windows 8 on domain "B". No trust relationship. When browsing to server with IE, it asks for username & password. After typing correct ones, login succeeds and web application works ok. So I can confirm that this fix seems correct (against my initial thoughts - sorry for that). |
|
Sounds like it's time for 1.7.4. I wasn't to happy with unit tests but sounds like fix is good. I'll be doing that off the 1.7.x branch so a little work involved to get it ready. I'll try to get this out later this week. --- Original Message --- From: "Ari Suutari" notifications@github.com Tested again with 1.7.4-SNAPSHOT (compiled today). Server is Win2012 on domain "A". Client is Windows 8 on domain "B". No trust relationship. When browsing to server with IE, it asks for username & password. After typing correct ones, login succeeds and web application works ok. So I can confirm that this fix seems correct (against my initial thoughts - sorry for that). Reply to this email directly or view it on GitHub: |
|
👍 |
A fix to capture tomcat 6, 7, and 8 negotiate authenticator Win32Exception which started occurring around June 2015 per issue #107. Around that time Exception was switched out for IOException which allowed Win32Exception to occur. Now catch that and perform same actions as IOException. Left original unit tests capturing fall through and added new check as this shows Negotiate, NTLM where the comma was normally expected to be a space. This does fix the condition but I'm not 100% certain this is correct from unit test side.
A fix to capture tomcat 6, 7, and 8 negotiate authenticator Win32Exception which started occurring around June 2015 per issue #107. Around that time Exception was switched out for IOException which allowed Win32Exception to occur. Now catch that and perform same actions as IOException. Left original unit tests capturing fall through and added new check as this shows Negotiate, NTLM where the comma was normally expected to be a space. This does fix the condition but I'm not 100% certain this is correct from unit test side.
A fix to capture tomcat 6, 7, and 8 negotiate authenticator Win32Exception which started occurring around June 2015 per issue #107. Around that time Exception was switched out for IOException which allowed Win32Exception to occur. Now catch that and perform same actions as IOException. Left original unit tests capturing fall through and added new check as this shows Negotiate, NTLM where the comma was normally expected to be a space. This does fix the condition but I'm not 100% certain this is correct from unit test side.
|
Ran out of time tonight but essentially I've got the 1.7.x branch ready for a push. Please ignore the 1.7.4 branch I created. It's just a copy of the master currently and I will remove it later. I need to review everything on the 1.7.x branch before I go ahead and push that and notify the group. I should be able to get this up tomorrow and have it available by Thursday in central. |
There appears to be a bug introduced in last week or so to the project affecting waffle-test and tomcat modules unit tests where 'the token supplied to the function is invalid'. This is a marker so that we don't release until this has been resolved. Sufficient logging exists around this to log the condition. More research is required to resolve this.
The text was updated successfully, but these errors were encountered: