[opensource-dev] Login response request/processing in 1.23 vs Snowglobe / 2.0

Kitty sldev at catznip.com
Mon Apr 5 14:22:00 PDT 2010


*waves*

What changed in the login response request/processing code from 1.23 ->
Snowglobe -> 2.0?

I've been trying to figure out why I can connect to the beta grid, but not
the main grid half the time with Snowglobe / 2.0, but can effortlessly with
1.23.

Alt logs on fine to both grids under any combination of viewer.

As best I can determine it's the time between authentication and finish
receiving the folder structure etc login response dump that's causing the
issues (alt has a much smaller inventory and beta grid has a much higher
transfer rate than the main grid)...

Main avie, main grid, v2.0:
2010-04-05T20:19:56Z INFO: Poller::Poller: login_to_simulator request sent
to https://login.agni.lindenlab.com/cgi-bin/login.cgi
2010-04-05T20:22:17Z INFO: LLXMLRPCTransaction::transferRate: Buffer size:
2235778 B
2010-04-05T20:22:17Z INFO: LLXMLRPCTransaction::transferRate: Transfer rate:
127.272 Kb/s
2010-04-05T20:22:17Z INFO: Poller::poll: login_to_simulator result from
https://login.agni.lindenlab.com/cgi-bin/login.cgi: status Complete,
errorcode OK ((done))

About 137 seconds, login halts on "Connecting to region" (sim has given up
waiting I'd guess)

Main avie, main grid, v1.23:
2010-04-05T20:31:33Z INFO: LLUserAuth::authenticate:
LLUserAuth::authenticate:
uri=https://login.agni.lindenlab.com/cgi-bin/login.cgi
2010-04-05T20:31:33Z INFO: LLStartUp::setStartupState: Startup state
changing from STATE_LOGIN_AUTHENTICATE to STATE_LOGIN_NO_DATA_YET
2010-04-05T20:31:41Z INFO: LLStartUp::setStartupState: Startup state
changing from STATE_LOGIN_NO_DATA_YET to STATE_LOGIN_DOWNLOADING
2010-04-05T20:31:56Z INFO: LLXMLRPCTransaction::transferRate: Buffer size:
2235778 B
2010-04-05T20:31:56Z INFO: LLXMLRPCTransaction::transferRate: Transfer rate:
106.384 Kb/s
2010-04-05T20:31:56Z INFO: LLUserAuth::authResponse: Processed response: 0

About 23 seconds, normal login.

So I'm wondering what changed to that bit of code to cause such a wide
difference in download speed? *confuzzled*

(Actually on a hunch I peeked at the forums and the above problem would seem
to fit quite a few of the stalls at "Connecting to region" that some people
are experiencing/reporting. In the cases where there are no logs posted
there's a report that logging on with 1.23 will always succeed and when
there is a log posted there's the same exceptionally long time for
downloading the login response dump)

Kitty



More information about the opensource-dev mailing list