On standalone, +1 for a silent logout of the hung session. No message at all, it seems to be the most sensible.
In grid mode, we should actually enforce the waiting time, and the new user server will have that as an option. This is because in grid mode, if you send a message to log out the user, the sim the user is in will log them out. Then, currently, we set the login status to not logged in, anticipating that response from the sim. However, if the sim is slow to log out the client, or misses the request, and the client logs into another sim by name, they will actually have 2 root agents i the grid. If they then teleport to the sim where the orphaned root is, that sim will crash. The new user server will address this by recording success/failure of the logout request. It will send a REST logout request to the client's last known sim. If that sim is reachable, it will then post the alert back to the user and will leave the logged in flag set. This will cause all subsequent login attempts to fail until the sim completes logout processing. If the sim is not reachable, the flag will be cleared immediately, assuming the sim is down and therefore holds no orphan root agent. Restarting a sim already clears the users that were recoded to be in that sim, so the approach seems robust. Melanie Arthur Valadares wrote: > Stefan, actually those 5 minutes for standalone are actually not true, > you can login right after it gave you the message to wait. So the only > thing we get today is a warning message that you're supposedly already > logged in, but the same symptoms apply (current user is kicked out of > region so new one can login). > > Melanie, the patch I applied is already for standalone only. I was > actually wondering about grid, but since you're saying it's needed, > won't even have to look at it. > > I think I will make this standard behavior, so we won't have any more > stuff on OpenSim.ini that we already have. > > Unless someone is strongly against for some reason :) > > Thanks all, > > On Sun, 2009-05-31 at 13:01 +0200, Stefan Andersson wrote: >> How about configurable, with 0 being immediately? >> >> Best regards, >> Stefan Andersson >> >> >> >> >> >> ______________________________________________________________________ >> Date: Sat, 30 May 2009 20:52:19 +0200 >> From: [email protected] >> To: [email protected] >> Subject: Re: [Opensim-dev] Is the message "Please wait 5 minutes" >> still needed? >> >> Frisby, Adam wrote: >> Make it standard. >> >> +1 >> >> +1, since the potential other session is kicked anyhow already... >> Adam >> >> >> -----Original Message----- >> From: [email protected] >> [mailto:opensim-dev- >> [email protected]] On Behalf Of Arthur Valadares >> Sent: Friday, 29 May 2009 2:59 PM >> To: OpenSim-Dev >> Subject: [Opensim-dev] Is the message "Please wait 5 >> minutes" still >> needed? >> >> Hi everyone, >> >> I wanted to consult you all about this matter, even though I >> have >> already committed some related code. I will gladly change it >> if we >> decide to do things differently. >> >> Once upon a time, there seemed to be a necessity that you >> had to wait 5 >> minutes if you crashed and tried to log back in. Those were >> the days >> before my time, tales that the elder tells us by the >> fireside. After >> these days, when I started messing around with OpenSim, >> there was a bug >> that would kick you after a few minutes if you tried to log >> back on >> without waiting. This was also fixed (thanks Crista!), which >> leads me >> to >> my question. Why are we still getting this message? >> >> Well, the only purpose I can think of is it warns you that >> you either >> crashed last time or that someone is using your account. >> Either way, >> something went wrong and the user might want to know that. >> >> So the code I committed puts an option in standalone that >> allows you to >> skip this message and just let OpenSim kick the old user and >> connect >> the >> new one, kind of like what happens in messengers if you >> logon from a >> different location. This is very useful when you have >> softwares that >> encapsulate the login process and doesn't want to leave the >> user >> looking >> at the viewer's login page with a failed message. >> >> So, should this become standard and just skip and kick, or >> should we >> leave it up to configuration? Opinions please.. >> >> >> -- >> Arthur Valadares <[email protected]> >> >> _______________________________________________ >> Opensim-dev mailing list >> [email protected] >> https://lists.berlios.de/mailman/listinfo/opensim-dev >> >> _______________________________________________ >> Opensim-dev mailing list >> [email protected] >> https://lists.berlios.de/mailman/listinfo/opensim-dev >> >> >> >> -- >> dr dirk husemann ---- math & computer science ---- ibm zurich research lab >> RL: [email protected] - +41 44 724 8573 - http://www.zurich.ibm.com/~hud/ >> SL: [email protected] --------------------- http://xyzzyxyzzy.net/ >> _______________________________________________ >> Opensim-dev mailing list >> [email protected] >> https://lists.berlios.de/mailman/listinfo/opensim-dev _______________________________________________ Opensim-dev mailing list [email protected] https://lists.berlios.de/mailman/listinfo/opensim-dev
