BNLS Problems

[Cyberpenguin]

AFAIK Stealthbot tries to connect to bnls.valhallalegends.com and if that does not work, it tries to connect to valhallalegends.com I also encountered this error yesterday, and i changed my valhallalegends address from bnls.valhallalegends.com to valhallalegends.com and it works fine


Conclusion, use valhallalegends.com as your BNLS server.

[Mephisto]

Okay, so BNLS is now able to accept connections and initiate the protocol with the client, but it's giving us invalid configurations which results in an error from Battle.net (0x101 on 0x51 return).  Skywing or Yoni?

To save them the trouble: What?  Post a dump of a failed negotiation, and check that it recurs more than once in a short period of time.  There's a known issue with BNCS version checking that if the version formula changes while you're computing your answer, you'll be incorrectly told "Wrong answer".  If BNLS was responding slowly while you were testing, that would increase the chance of running across this issue.

It was my invalid code; my apologies.  But interestingly enough I came accross another bug: awhile ago I implemented the WAR3 connection w/ BNLS -- BNLS gave me an incorrect password proof and thus the complete logon failed (though I could logon I could not pass the proof); but when I used JBLS everything worked fine.  And I tripple-checked everything in my code and did exactly as is documented on the Battle.net and BNLS protocols.

Note: If JBLS does not follow the BNLS protocol and do exactly it 100% then this error may not apply to BNLS; but interestingly enough it worked with JBLS and not BNLS.

[Blaze]

Conclusion, use valhallalegends.com as your BNLS server.

Thats a bad idea. Thats the server that stealth used from stealthbot in 2.4R3 and all of a sudden it went down. Do you know how many idiots were spamming us  at stealthbot.net with error reports about this? www.valhallalegends.com:9367 is not the official bnls server. Use bnls.valhallalegends.com:9367.

[Adron]

It was my invalid code; my apologies.  But interestingly enough I came accross another bug: awhile ago I implemented the WAR3 connection w/ BNLS -- BNLS gave me an incorrect password proof and thus the complete logon failed (though I could logon I could not pass the proof); but when I used JBLS everything worked fine.  And I tripple-checked everything in my code and did exactly as is documented on the Battle.net and BNLS protocols.

If this happened only once or twice, it might be a bug in battle.net. Ask Skywing about it.

[tA-Kane]

I too had this problem, and did a little investigation during the experience. It seemed to me that ns1.vpnsys.org went offline for one reason or another. ns1.vpnsys.org is the main nameserver for a fair amount of the valhallalegends.com domain, including bnls.valhallalegends.com. So, if the nameserver was down, no one could convert the domain bnls.vallhallalegends.com to the proper IP address. The biggest (but not the only) exceptions would be people who still had the domain in their domain resolution cache, or who were using a nameserver which also had the domain resolution in its cache.

[Mephisto]

It was my invalid code; my apologies.  But interestingly enough I came accross another bug: awhile ago I implemented the WAR3 connection w/ BNLS -- BNLS gave me an incorrect password proof and thus the complete logon failed (though I could logon I could not pass the proof); but when I used JBLS everything worked fine.  And I tripple-checked everything in my code and did exactly as is documented on the Battle.net and BNLS protocols.

If this happened only once or twice, it might be a bug in battle.net. Ask Skywing about it.

It's an ongoing problem.  I tried it on different days, different times of the day, 10 times in a row sometimes, etc.  The major difference of JBLS and BNLS is that JBLS uses iago's hashing algorithms and BNLS uses Skywing's so it leads me to believe that Skywing's hashing algorithm produces errors perhaps?  But I doubt this since other's have successfully connted with WAR3 using BNLS.

[iago]

Although I didn't read this thread, I think the problem was with dns lookups.  I wasn't able to resolve vL's ip for most of yesterday.  Coincidentally, Thing's nameserver blew up recently.  I would imagine it was just a DNS burp.

[The-FooL]

Isn't this bad though because we don't know who may be hosting it. I never specified for Stealth Bot to connect to any other server accept BNLS. JBLS sends everything plain-text and if it wanted could log cdkeys/passwords. I am not saying that person will. I just think users should know where they are connecting to...

Stealthbot does not use a JBLS server(at least by default).  And, there are no public JBLS servers up, that I know of.

[Skywing]

It was my invalid code; my apologies.  But interestingly enough I came accross another bug: awhile ago I implemented the WAR3 connection w/ BNLS -- BNLS gave me an incorrect password proof and thus the complete logon failed (though I could logon I could not pass the proof); but when I used JBLS everything worked fine.  And I tripple-checked everything in my code and did exactly as is documented on the Battle.net and BNLS protocols.

If this happened only once or twice, it might be a bug in battle.net. Ask Skywing about it.

It's an ongoing problem.  I tried it on different days, different times of the day, 10 times in a row sometimes, etc.  The major difference of JBLS and BNLS is that JBLS uses iago's hashing algorithms and BNLS uses Skywing's so it leads me to believe that Skywing's hashing algorithm produces errors perhaps?  But I doubt this since other's have successfully connted with WAR3 using BNLS.

There is a buffer handling bug in Battle.net's checking of "NLS"-style passwords that causes it to incorrectly fail a small percentage of logons.  It's actually possible to detect when this will happen clientside, but this can be done so late in the logon process that the only recourse is to abort the connection and retry the whole process again, hoping the server doesn't pick values that make the buffer handling bug visible.

At some point, BNLS will be modified to close connections if somebody requests a logon with values that will cause the server to report an incorrect result.

[tA-Kane]

At some point, BNLS will be modified to close connections if somebody requests a logon with values that will cause the server to report an incorrect result.

Eww! BNLS should report to client that the data will cause the server to report an incorrect result! Let's not force the client to reconnect to both Battle.net and BNLS, but just Battle.net.

[Skywing]

It would be preferable to extend support to all programs, not those updated after the change goes into effect.  I'll see about making it possible to bypass a disconnect if the client specifies some knowledge of an enhanced protocol that has some type of status code.