TPing from Thera to Mohrr is one example.. On windlight I get it every time i try to tp today, while yesterday (before the 2000 region restarts) it was tping ok.

On the RC i think i did manage to TP successfully from mohrr to thera so its not every time there

After i try to teleport wither using map or lm, the bar moves slowly at first, and after a bit i get the log out message:
You have been logged out of Seconde Life:
You have been disconnected from the region you werre in.
Click continue blablabla (the casual [continue] and [quit] buttons)

Happened yesterday 4 times in a row for me with different TP targets.

Repro: TP from "Isle of Ecstasy" to "ETD Isle (212, 216, 26)" via Landmark
Isle is up and running (11 ppl on it (reported by "Map"))
Date/Time: 17.11.2007, 3:26

After logging back in my AV is stuck (unable to move).
A log out and new login results in a "System is logging you out right now". Which means a real working login after such a failed TP takes me often 2-5 minutes.

For me this seems to be linked with 'missing groups' where I have no groups in my group editor, presumably because some lame server somewhere can't be bothered to download them .. which I notice fairly quickly because I don't have rights to any of my stuff and can't manage my land. Anyway .. I''ve started watching this now .. I haven't yet seen a case where I get this sort of teleport failure without groups also being borked. It also seems predictive .. if my groups are borked I know the tp will crash me. Anyone else noticing that link?

i get the message of being logged out only when my groups list hasn't loaded.
that resolved the randomicity confusion from before.

When groups list is complete, TP's happen normaly

I thought it might be more prone to occur after being in-world for a long time (e.g. due to memory leaks accumulating) but it just this morning occurred again (11/19/07 Monday) after being in-world for only an hour or less.

PS.This problem might not be VWR problem but something else?

I've had this problem too, but not yet have a repro.

Have been home bound for over a week now, have given up trying to tp

Am running a G5 Mac, running OSX 10.4.11, can anyone help

This morning's incident occurred with my Group List fully populated and intact – I happened to have looked at that shortly before TP'ing to my doom.

.. I first noticed it when the server patch to add teleport debugging was added .. October sometime? But since then it has gone from an occasional nuisance (1 time in 10) to a major headache (whole days without a single successful tp). It seems load related ... much less likely to happen when SL is quiet (looking forward to TG weekend), but one has to mistrust that sort of anecdotal evidence.

.. my hardware, router, and router config have been stable for 9+ months now. Since I am running a microsoft OS I can hardly use the word stable .. but you get the idea.

i) I can agree that the tp failure/logout is not invariantly associated with unpopulated group list. But in my case at least, the correlation is 98%+ so I can't rid myself of the notion that they are connected.

ii) It also afflicts intra-region teleports via the map. So a region -change is also not necessary to produce this crash mode .. although I don't know how intra-region tps are coded .. possibly they use exactly the same mechanism in which case this might not be a very interesting observation.

If I stay in one place for a long time and then teleport, the teleport fails.
If I log back on and teleport right away and TP regularly from one place to another, the teleports work.

Symptom of failure:

• the viewer image goes monochrome / grayscale
• a message box appears saying:

"You have been logged out of Second Life:

You have been disconnected from the region you were in.

Click Continue to look at existing IM and Chat.
You will not be able to perform any other operations.
Click Quit to exit Second Life immediately."

with buttons Continue and Quit

It's unfortunate that this does not trigger a crash report. So, for the record this is a 1 year old iMac with Core 2 Duo, nVidia 7600, OS X 10.4.10 and running SL1.18.4(3).

You are at 271702.7, 259462.3, 28.3 in Avaron located at sim4899.agni.lindenlab.com (64.129.47.153:13003)
Second Life Server 1.18.5.73200

CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 6000+
Memory: 3033 MB
OS Version: Linux 2.6.22.9-61.fc6 #1 SMP Thu Sep 27 18:48:03 EDT 2007 i686
Graphics Card Vendor: NVIDIA Corporation
Graphics Card: GeForce 7300 GT/PCI/SSE2/3DNOW!
OpenGL Version: 2.1.1 NVIDIA 100.14.19
LLMozLib Version: 1.1.0 (Mozilla GRE:1.8.0.12_0000000000)
Packets Lost: 5285/140277 (3.8%)
Viewer Digest: de123326-c0a8-4b7f-daa1-afb79e8c2d45
----------------

2007-11-26 12:53 PST (SL Time) Teleport Warbluster -> Isles of Intrigue2
2007-11-26 14:04 PST (SL Time) Teleport Isles of Intrigue2 -> Avaron

Logging in after failed TPs most commonly returns me to my set Home position, rather than either the TP source or destination.

Symptom of failure:

• the viewer image goes monochrome / grayscale
• a message box appears saying:

"You have been logged out of Second Life:

You have been disconnected from the region you were in.

Click Continue to look at existing IM and Chat.
You will not be able to perform any other operations.
Click Quit to exit Second Life immediately."

with buttons Continue and Quit

Just for the sake of it I'm on a Toshiba laptop with an Intel 945GM, ! GB RAM

So, unfortunately not a lot of rhyme or reason to it, not reproducible on demand, but quite annoying just the same.

Symptom of failure:

• the viewer image goes monochrome / grayscale
• a message box appears saying:

"You have been logged out of Second Life:

You have been disconnected from the region you were in.

Click Continue to look at existing IM and Chat.
You will not be able to perform any other operations.
Click Quit to exit Second Life immediately."

with buttons Continue and Quit

If you notice, this issue occur when you change between regions... not in the same region. If you walk, fly or tp into another region... crash!!!!
This is happen to me since 1.18.4(3)

Sometimes TP works as somebody above said... if you do it very quick... but this success not happen very often and sometimes can not be replicable. My computer is an iMAC power PC.
I have an adsl modem with 2Mbps of bw.... If the TCP or UDP ports are the problem, you need to tell us how can we configure our firewall options...
pls help!

However, I do hope the mindset is not, "it doesn't affect that many people, let's ignore it".

For those who are affected it can be critical in its impact. In my own case it affects me about 90% of the time .. making it essentially impossible to use SL.

This inconsistency is one of the more puzzling features.

90% of Residents seem unaffected.

9% are affected infrequently.

<1% see this behaviour almost all the time.

No one (that I am aware of) sees this bug all the time.

Oh .. and if any Linden browsing here maintains the illusion that the support ticket system works .. I invite them to look at the file.
My ticket has been open for two-and-a-half weeks. I've yet to receive a response from anyone who manages to convince me that they have even read the ticket.
That sort of support actually has negative impact.

I don't think your estimate is accurate, it seems to happen to a lot of people. More likely those people just don't bother to find this entry on the JIRA, or even mention it at all. Flaky teleport's have become something of a standard SL undocumented feature over time. Either leaving you where you started frozen or able to try again, or crashing you out, or sometimes teleporting you and then crashing just for variety.

The recent upsurge in the frequency of this particular flavour of crash happening is a worry, but I honestly think a lot of SL users have become desensitised to bugs. "Yeah, I've had about X crashes already tonight" is a fairly common comment if you complain about your last crash in an area with more than about 5 people present.

Ahh btw I forgot to mention that this bug is still happening, maybe not so frequent like before (maybe also related to the cache that i have to clear now daily), but i still suffer from it :/

I am not really sure, if I alone would easily produce an excess of ten crash reports per day, since weeks, and all the other people who are suffering the same fate would duly and automatically flood the crash reports in at the same rate, if it still would have been downgraded. Or if it might have even been resolved by now.

Well, one might still dream, might one not...

It's so bad for me that I just can't go anywhere... almost any time I try to teleport I simply get logged out and rolled back. I'm pulling my hair out. I will provide two useful bits of information I've learned from this for others suffering from the issue:
1: if you do try to teleport and don't see any status text by the time the bar is about a quarter of the way filled, you can hit cancel and send your friends IMs telling them you're about to be disconnected so they don't wonder why you logged off. But by this time most people just assume you got TP-booted.
2: A better idea is to forget teleports completely and simply log off, set your starting region to wherever you want to go, and log back in. Right now it's the only way some of us can get around.

I'll have a few tests but it's likely I'll be removing my vote if my problems appear to remain random and not sim specific.

Please don't jump to conclusion based on your isolated experiences and please don't jerk other people's bug reportage around.

If you have really hard, non-subjective evidence of region specificity then you need a new strand or a new bug report, but please don't try and invalidate everything we have collected here.

This not region specifi, it happens everywhere, the only thing i noticed (someone else already reported it), is that it happens always if i stay in a region for "some" time (approx more than 20 minutes), then the TP's stops to work... but if i try multiple teleports just after have logged in, they works correctly.

I've limited knownledge about programming, so i may only guess why the time factor is gonna affect the TP, it's something like a buffer (because of a memory leak or any other similar problem) is going to get full after a certain period... so the tp's stops to work.

Some people get logged out every time, no matter where they are. Some others get logged out consistently when leaving specific sims.

All I know is that if you hang out in the Plio region for a few minutes, and then try teleport/map off... you will get logged off most times, though not every time. I do not have a reliable repro, but I run a business there that relies on a large number people coming and going, and have heard this from several people. Its consistent.

And if there is a specific, reproducible issue TPing to a certain region(s), as Fluf Fredricksson mentioned, please file it as a separate issue (and you can link it to this as related). It's important we be able to focus on this issue without diluting it.

I (and several other Lindens) have noticed this on occasion but it doesn't happen frequently to me. But because of the votes and newfound reports (which give us a better idea of how many are affected), I'd agree this is a Critical issue, so I'm moving that back up.

The summary is now vaguer and not as useful as my earlier edit, so I'm going to revert it.

Those of you saying getting logged out happens "every time" you teleport, please be sure to attach a log file so we have further info to go on – info on how to do that is here:

» http://wiki.secondlife.com/wiki/Debug_Help

We'll continue to investigate, thanks for your patience.

More observations:
The incidence of this for me has been dramatically reduced since the last rolling restart on the 28th/29th? of November. It was pretty much a permanent thing with occasional spells of normal operation up until then. Since then I have had only one or two bad days.
For me it is still strongly linked (99%+) with unpopulated groups.
If I attempt to abort the failing teleport early I sometimes get the "Teleport is currently blocked" error THEN get logged out.
For me definitely not location sensitive ..while in this mode all teleports from/to anywhere by any method will fail and log me out.

Log files attached after having been lightly edited to remove personal data.

***Folks .. before you post your log files to a public place .. be aware .. they contain your user name on your workstation. *** (bad Lindens, bad)
***and the name you logged in under .. which may not be the name you logged into JIRA under .. so if you are sensitive about having your alts linked ..... *** (really bad Lindens)

And having said all that: I also noticed a relevant drop in the amount of failures to teleport properly since the latest update outage.
Nevertheless it still happened...

Background:
Until (and including) 1.18.3, everything worked fine. I missed 1.18.4, but with the WL-clients, and all the ones since 1.18.5 (FL and official) I got logged out on TP/sim-crossing.

Changes:
Between 1.18.3 and 1.18.5, library c-ares was introduced for doing DNS lookups.

Problem:

• Network trace revealed that a faulty configured (local) DNS-cache delayed the answer for a DNS request for about 5s (it first asks the wrong
  DNS server for the name, and after a 5s timeout the correct DNS-server, so every DNS-request took just over 5s).
• c-ares default configuration uses a timeout of 5s.

So, what happened was:

• DNS request for the address of the sim I want to TP to is posted via c-ares
• DNS-cache asks wrong server first
• after 5s, c-ares timout hits and c-ares asks again, causing the DNS-cache to drop the old request (from the same IP, for the same name)
  and restart the process, asking again the wrong DNS server.
• after several retries, c-ares tells SL-client: Can't find address of sim
• client logs off.

Solution:

• For me, correcting the faulty DNS-cache configuration solved my problem. DNS requests are now answered within <0.2s, and I can TP with the current 1.18.6 (and
  the 1.18.5 I compained about above) without fail.
• In general: This will happen whenever the answer for a DNS-request is delayed for more than 5s (neatly explaining the "I'm sometimes logged out" problems, too).
  I suggest to configure c-ares with a higher DNS-timeout (at least 15s, via ares_init_options). Will add a patch later (have to switch to the 1.18.6 sources first).

Nevertheless, I just tried it again to make sure: Fixing the DNS setup fixed the problem for me. I can get the old non-TPable condition again by using the old config for my DNS-cacne.

So, to summarize Slow:DNS responses were the reason for the problem for me, but the introduction of c-ares into the viewer at the same time the problem occurred for me seems to be a coincidence.

However, I'm fairly convinced it's some kind of configuration issue, since the code itself hasn't changed significantly in a while.

The client is being logged off because the sim you are teleporting from is telling the client "we are done". However, the sim you are teleporting to has not told the client "Yeah, you are here now".

Basically the existing region information within the client is being removed, but there is no new region information, and as such the client immediately thinks you are logged off, hence seeing the Logged Off screen.

Now the interesting part here is that if this were a configuration issue within sim to sim communication, then teleporting wouldn't work at all. And yet it does. Add to that accounts where this happen, I then hijack them and it all works fine for me, and we arrive at a sim->client configuration issue that may well be related to local conditions at the client. Something is not getting forwarded correctly - potentially the new sim saying the client "hey, I have you". And it's very intermittent.

I'm still looking into this. Just thought I'd update you guys on what I am seeing.

Umm .. can I just reemphasize that 'intermittent' is only the word to use if you are averaging across the totality of Resident experiences. For certain individual Residents it is certainly not intermittent. The ability to tp at all without a crash is the intermittent bit ;-(

Having said which .. and for what little it's worth .. some further observations.

• The incidence has changed since the last grid-wide rolling restart of about ten days ago. It's dropped from perhaps 95% of all tp attempts (bursty - ie that 95% means one hour a day when things work properly not 1 out of 20 attempts) to perhaps 70% of the time when it's broken.

• The (for me) very strong co-incidence with unpopulated group lists has weakened slightly .. dropping from 99%+ to 95%+. I still never see a tp succeed with unpopulated groups .. but I do now see the occasional one fail even with populated groups.

• Watching myself in a failed tp using an alt I am often seen to 'arrive' briefly at my destination before vanishing again as I am logged out. From my perspective I never see my destination. That tends to support Jake's analysis I imagine. The new region 'receives' you but the viewer doesn't get the news.

• Cancelling a 'failing' tp is usually fatal in the sense that you end up being logged out anyway. However you sometimes see strange things in the second or two before you are logged out. The one that intrigues me is a curious 'mixture' of origin and destination. Once or twice this has even extended to the view with things like objects from the origin against the landscape of the destination. The commonest though are things like a scrambled title bar .. name of destination region with description from the source region is quite a common one. Anyway .. the viewer is getting some information from the destination.

Dunno if any of that helps ... but omg I hope something does and soon.

Gil, if you are in world, I'll give you a ping.

More Information:
Internally, Linden Lab is in the process of bringing up a new internal second life grid, and funnily enough any teleport from one location to another fails, with almost exactly the same symptoms. I am hoping that whatever is the issue there is the same one that is being seen in the main second life grid.

So fingers crossed I find that and the solution is applicable.

Pls let us know as soon as your team finds something!

Enable TCP port 12043 through your router.

I was able to repro/resolve this issue consistently by opening/closing this port. This also solves the missing group issue here -> http://jira.secondlife.com/browse/VWR-2633.

I'm fairly sure I have the area where this is happening narrowed down. I fixed our internal problem and it suggested the area where this happens externally.

I also suspect this is potentially a router / firewall issue, but I need some extra data to add evidence to my hypothesis.

With that in mind, could I ask for some more data when you guys run into this? Namely -

a) Can you region cross when the Teleport goes RART**? IE can you walk / fly from one region to another?

b) Does voice still work when you are in this state?

c) Is this a state that persists across multiple logging in?

d) does it happen instantly on logging in, or is this something that starts occurring over time?

e) What router do you use?

f) What firewall do you use?

g) Does it go away if you power cycle your router

and finally

g) What deoderant do you use? Cos I get a rash from mine...

If this is what I think it is I'm gonna have a hard time proving it and an even harder time figuring out what to do about....

• • RART = Rapidly Assuming Room Temperature. Emergency room code for Dead.

b) Does voice still work when you are in this state? NO IDEA. THERE'S VOICE NOW?

c) Is this a state that persists across multiple logging in? YES

d) does it happen instantly on logging in, or is this something that starts occurring over time? EITHER. BUT IT DOESN'T "START OCCURRING" .. IT'S ON/OFF WITH A TIMESCALE OF HOURS.

e) What router do you use? USR SURECONNECT

f) What firewall do you use? The USR SURECONNECT INTERNAL ONE. WINDOWS SO-CALLED FIREWALL COMPLETELY DISABLED

g) Does it go away if you power cycle your router HAVEN'T TRIED. THAT WOULD TAKE A LONG TIME TO ESTABLISH WITH ANY CONFIDENCE.

and finally

g) What deoderant do you use? Cos I get a rash from mine...

I am on a Zoom ADSL X6 modem router, connecting via wireless to the router. I have Vista Firewall on but no other firewall. I think there is also a hardware one in the router. I cannot answer most of your other questions because the tp failure caught me unawares so I was not testing, but I tend with that alt to be tping everywhere (mostly on private land so region walking is not really on most of the time). Never got voice to work (I think my headphone set is broken rather than SL).

If you are worried about deodorant, ask for a transfer to the Brighton office, its never warm enough in England to need deodorant.

a)NO
b)i dont use voice
c)YES
d)both
e)2WIRE
f)above....
g)no

If i can give you more data, pls dont hesitate in ask for it.... This situation is unsustainable

And can you share your theory with us? I'm having trouble buying the router theory because of way the fault comes and goes, it has to - at least - be much more elaborate than just router configuration. And that wouldn't answer the question of why it just suddenly started occurring .. at the very least something at your end must have happened to expose it.

I could buy a routing connection .. which would explain why the problem seems to occur with all accounts from a specific machine but not with those accounts used on a different machine. The route used between here and there could well change .. over a timescale of hours .,, and produce the pattern seen.

I'd be fascinated to know, for instance, whether a specific route to LL's datacentre makes me more or less likely to be connected to a particular server/rack/interconnect ...

Or a variation of Homer's theory .. a problem only exposed by high latency .. which could also vary on those long timescales ...

b) Does voice still work when you are in this state?
R: Not sure.

c) Is this a state that persists across multiple logging in?
R: Nope, immediatly after login the TP's works correctly. The problem appear after a certain amount of time.

d) does it happen instantly on logging in, or is this something that starts occurring over time?
R: as c) ...it happens only after a certain amount of time sitting in a sim.

e) What router do you use?
R: netgear... but it's the same router as i had before the problems stard (about a month ago?)

f) What firewall do you use?
R: router firewall.. but as e) the configuration didn't changed.

g) Does it go away if you power cycle your router
R: it's not necessary, since at the next login TP's works again.

Anyway the effect (exposed by jake) seems right: the avatar reaches the destination, but the client doesn't receive the information so it assumes it is a failing TP and try to recover the originating simulator, but it has been "cleared" (since the TP, in fact, worked) and then you're forced to logoff.

Im not quiet sure what i have now... but let me tell you about it....
Yesterday i was trying to teleport myself (always try to "check" the issue....) so... when i saw the progress bar had that delay i sent SL app to background and i pull up the SAFARI app to the foreground and stared to navigate in the internet.... and after a couple of seconds i heard the "bell" yes that bell that all of you heard when the teleporting action is completed.....
My surprise was i did a sucesfull teleporting!!!! yeah after a couple of weeks!!!!
I realized that the only diference was that i was surfing the internet at the same time when i was teleporting myself.... so I replicate the scene and i teleport 4 times before SL takes me out of the game.... with the same error window....
That was yesterday.... today i did just 2 succesful teleports..... same scenario...

SL in the background
Internet browser in the foreground.... changing pages....

Peharps did means nothing for an unexpert user as myself.... i hope JAKE or else more expert in these stuff can figure out was happened....

Guys.... pls think hard... seems you are close enough to make a workaround or some....

Happy New Year!!!!

So here's what I think this is.

Normal communication between the Sim and the Client is via a UDP connection that is hand managed by our code.

However, recently we've built a better way of dealing with some of these communications - since they don't need to be UDP (the difference between UDP and TCP/IP - which I'm sure you've all heard of - is that TCP/IP guarantees packets delivered in order, and actually delivered. UDP does not. It's like sending a letter and just throwing it out into the world and hoping it gets there, and having registered mail. So why don't we use TCP/Ip purely? Because if a packet gets lost or is delivered out of order then TCP/IP can take on the order of seconds to sort it out and get the right packets where they are supposed to go. Something you can't afford to rely on for things like positional updates and so on).

Some communications can afford to be TCP/IP since it isn't so critical that we update instantly - stuff like inventory handling, IM's and so on.

Anyway, the new structure we have in place is also an indirection system. What that means is that instead of the client assuming that it's always talking with a Sim, on log on the sim tells the client "this is the web address you talk to to handle IM's. This is the web address you talk to to handle Inventory. This is the web address you talk to to handle Asset downloading" etc. The idea being that eventually we can move the actual processing of inventory, or IM's or whatever off the sim itself, and onto a dedicated system instead. This frees up processing time on the sim to handle physics and scripts, and allows us to scale the way we handle IM's or some database intensive task.

So inside of the teleport (and region crossing) there is One Message that goes via this new communications method (called Capabilities). All the rest are UDP, however one goes via TCP/IP and is sent via a web connection rather than anything else. If that one message is dropped then the teleport (or region crossing) fails, and you get logged out. But you only get logged on on that one specific message being dropped. If any other message is dropped you get the teleport failed message.

So, ergo, this means that one message is, in these cases, not getting through.

Now it should be further pointed out that this new communications method (capabilities) operates via HTTPS not just the standard HTTP protocol. HTTPS is 'secure', and we need that because we'll probably be passing some sensitive data across. However this does require the correct certificates in place (you've all seen websites that require you to accept a certificate right? That's HTTPS in action) and it's all a lot more complicated that traditional HTTP web communication.

What does this mean? I have a suspicion that in some cases some routers get messed up with HTTPS communication, over time. Some internal buffer gets overloaded, or some certificate gets messed up or whatever, with the net result of this one message not being communicated from the Sim to the Client and teleports suddenly failing.

Now in the short term the answer is obvious - switch the teleport / region cross message across back to UDP and all should be good again. But since we've built this new communications system, and a lot of our intended new sim configuration design for 2008 and beyond is dependent on this working, we need to understand why we are seeing this hiccup, and what we can do about it.

Thats why I am asking about routers and rebooting to see if the problem goes away, because if it does then it's obviously router related. The fact that only a very small subset of you are hitting this makes me suspect a local configuration problem rather than our code per-se because if it was our code, everyone would be seeing this problem (as they did on our internal grid I was helping to bring up).

I hope this gives you an idea of what I am looking at and why.

Jake

This is now very plausible .. but don't lose sight of other possibilities. Once you pin the blame on HTTPS glitches (for which you adduce strong circumstantial evidence) you also have to think about the local machine. I'd be just as inclined to suspect a bug in the TCP stack in the local machine or even idiocy by some other HTTPS client doing something horrible (I, for one, almost always have a browser running at the same time as SL). Against that .. rebooting the local machine doesn't seem to help. So still puzzling however you look at it.

I wouldn't be ready to rule out timeouts as a culprit either. We do sometimes see horrific levels of both latency and packet loss on SL. Remember that TCP doesn't guarantee delivery - it just guarantees to keep trying. Presumably something somewhere in the server will eventually decide that this is just not happening.

But .. you still need code changes, Jake. Your code needs to attempt to recover from this sort of hiccup .. not just give up and boot the viewer. At the most minimal end of barely acceptable .. take the viewer back to the login* page with the SLURL of the failed TP target already entered.

[* Which is also a puzzle. Presumably the same sort of HTTPS transaction occurs during a login. Yet they work. I speculate because the login logic doesn't give up straightaway if the HTTPS transaction fails .. it goes round again and has another go.]

If the problem is related to the caps event queue during the teleport, then perhaps some sort of error recovery needs to be implemented as Gil suggest.

thank you for the information you provided. I am anyway not sure that it is really related to a router problem.

When the idea with the router related problem came up I tried several things to have a workaround:
(the diconnect effect after/during TP happens to me occasionally)

a) I have this problem occassionally with 2 different local routers
b) I have this problem also without a local router (direct DSL connection)
c) Usually when it happens I have had several (1-4) successful TPs and am inworld for at usually at least 30 minutes before it happens the first time.
d) When it happens and I get disconnected and log back in, I am stalled until I relog again: I can not TP anymore and can not move or fly anymore, but I can do everything else (transactions, IM, ...).
After your info I reset my router before I logged back in: No difference ... stalled again until the next relog.

> The fact that only a very small subset of you are hitting this [...]
e) A lot more people told me they have the same problem - they just don't believe adding their comment/note to jira will help fixing it. They just live with it, sigh and wait for it to be fixed - they are simply frustrated.

But back to "on topic" about the only thing I might have changed concerning this is upgrading my SSL related libraries in Debian Linux. I didn't think SL used those though, so remain somewhat unconvinced. But for the record am now on:
libcurl3 7.17.1-1
libssl 0.9.8g-3
openssl 0.9.8g-3
And teleports have been good for a while. [Shrugs]
I'm also using a heavily patched version of 1.18.5.3 which may have more to do with it.

Edit: Just jumped in world. Asked first online friend I could IM if they get logged out on teleports. Answer? "About once every three times"....

I've taken some time to experiment. Here's a little info I have which I can confirm:

1: Though I get seemingly random disconnects as well, if I remain in one sim for too long, particularly overnight, I am guaranteed to have a disconnect when I try to move to another sim in the morning. This is 100% reproducible so far. It's possible that this particular reproducible case of TP disconnect is caused by being logged in for too long, or being in one sim for too long... I haven't tried to figure out which of the two causes this.

2: I live on an island, and WALKING to the adjoining island will fail as well. I have no idea if this happens on the mainland, but remaining on my home island overnight and trying to walk to the adjoining one in the morning is a 100% reproducible failure (a hang on sim crossing) just like an attempt to teleport under the same conditions will cause me to disconnect. Could it be the same bug? It seems likely, and there have always been people in the destination sim who aren't experiencing any lag or other issues there.

3: This doesn't mean it ONLY happens when I've been in one sim for a long time. I have, in one circumstance, logged in after being TP disconnected only to find I'd been rolled back to a different sim. I immediately tried to teleport and got disconnected. This happened three times in a row before I decided to simply log in at the sim I was attempting to teleport to. However, this is now rare.

4: Like others, I also often find I'm "stuck" upon login after a TP disconnect. My avatar is stuck in a ruthy crouch pose, I can not move, and can not teleport or take other actions. Another relog generally fixes this. However, this also doesn't happen 100% of the time.

5: I've sometimes played with logging in two avatars at once on the same system. (using the -multiple parameter.) When one avatar is TP disconnected, the other is entirely unaffected, and no other downloads or connections are interrupted.

6: I've removed my router entirely and dialed in directly with my DSL, and the results are the same. This is absolutely not the only cause of these disconnects. Whether it makes the problem worse or not, I can't say, but I can still reproduce disconnects without the router. However if it helps the process at all, my router is simply a linux box.

7: The failure in TP disconnects seems to happen in a consistent way. I know when I'm about to be TP disconnected because "Requesting teleport" briefly flashes and then disappears. The progress bar continues to move, but "contacting new region" never appears. Once the progress bar reaches the halfway point, I then get disconnected.

8: If I want, I can press the "cancel" button at any time during a failed teleport. I will then find myself at seemingly no location in the (origin) sim... the location vector does not appear at all at the top of the screen, but the sim name does. I am **NOT** truly disconnected at the time, though... I can send and receive IMs. Some time after this, at the same time I assume that the TP progress bar would reach the halfway point with no REAL teleport progress, I am then disconnected, just like I would be if I hadn't hit the "cancel" button. (EDIT: Just to clarify, the sim I'm stuck in is my origin sim, not the destination. As far as I can tell, SL never contacts the destination sim.)

I realize that it's difficult to track the real source of this problem, but I assure you it's not simply something that goes away when a router is reset or removed. Hopefully this information will help.

What makes me wonder is why a sim restart gets rid off this problem and not a router reset. Maybe it is burried again in some code lines for the server.
I had 6 tps of death from my sim to others, after a restart it was working again, so i would suspect also a different combination of the upper mentioned ideas.
And believe me, if a router messes up https communications, then throw it away. Otherwise you wouldn't be able to do other important secured web things.

b) Does voice still work when you are in this state?
R: Yes and the voice i hear is from the starting location (not the destination).

I've seen this behavior with the following clients:

• FirstLook 1.18.6 (76453) under Linux (nvidia 7600 GT, 2Gb Ram, Suse 10.3)
• Standard client 1.18.5.3 under Windows (Intel 945M, 1Gb Ram, Windows XP)

The Windows client crashes as often as the Linux client. (Which is only at TP-ing, other functions are very stable).

Dump of the log at the time of crashing (Linux-client):
-----------------------
2008-01-02T16:29:46Z INFO: removeRegion: Removing region 256768:259840
2008-01-02T16:29:46Z INFO: LLEventPollResponder::stop <9> https://sim4171.agni.lindenlab.com:12043/cap/23236922-505d-f6f4-450b-7f4b8cc1903a
2008-01-02T16:29:46Z INFO: LLMessageSystem::disableCircuit for 63.210.157.75:12035
2008-01-02T16:29:46Z WARNING: disableCircuit: Couldn't find circuit code for 63.210.157.75:12035
2008-01-02T16:29:46Z INFO: removeRegion: Removing region 257280:259840
2008-01-02T16:29:46Z INFO: LLEventPollResponder::stop <10> https://sim47.agni.lindenlab.com:12043/cap/b8460c8d-e679-67fa-5088-0092e34d6bef
2008-01-02T16:29:46Z INFO: LLMessageSystem::disableCircuit for 64.154.220.14:12035
2008-01-02T16:29:46Z WARNING: disableCircuit: Couldn't find circuit code for 64.154.220.14:12035
2008-01-02T16:29:46Z INFO: removeRegion: Removing region 256768:260096
2008-01-02T16:29:46Z INFO: LLMessageSystem::disableCircuit for 8.2.32.201:12035
2008-01-02T16:29:46Z WARNING: disableCircuit: Couldn't find circuit code for 8.2.32.201:12035
2008-01-02T16:29:47Z WARNING: removeRegion: RegionDump: Agapema 63.210.156.193:13005 { 257024, 259840, 0 }
2008-01-02T16:29:47Z WARNING: removeRegion: RegionDump: Neumoegen 63.210.156.135:13005 { 257280, 260096, 0 }
2008-01-02T16:29:47Z WARNING: removeRegion: RegionDump: Calleta 63.210.157.197:13006 { 257280, 259584, 0 }
2008-01-02T16:29:47Z WARNING: removeRegion: Agent position global { 257154, 260004, 502.846 } agent { 129.87, 164.313, 502.846 }
2008-01-02T16:29:47Z WARNING: removeRegion: Regions visited 2
2008-01-02T16:29:47Z WARNING: removeRegion: gFrameTimeSeconds 29826.7
2008-01-02T16:29:47Z WARNING: removeRegion: Disabling region Agapema that agent is in!
2008-01-02T16:29:47Z WARNING: createXml: Alert: [YouHaveBeenLoggedOut]
----------------------

Regards,
Marcoh Larsen

I'm pretty convinced that this is the message dropping, the question really is how is this happening.

The router issue - the reason I ask is that I was using a guy who commented in this thread as a control - he was having it happen every time he logged on, however when I hijacked his account it never happened for me. I give him his account back again and viola, it starts happening again - therefore the issue had to be down the chain from the Sim itself, and more likely a local issue. He then went out a purchased a new router and suddenly all his issues just went away. So you can imagine why I ask the router questions.

Now it appears obvious that his was a rolling red herring, although it is still some pretty strong evidence.

ISP issues is an interesting idea. Not entirely sure how I test that though.

This is the most interesting comment though
"What makes me wonder is why a sim restart gets rid off this problem and not a router reset."

Now that does make it interesting, since it sounds like the HTTPS message is being dropped from the sim to the HTTPS forwarding service which sends it on to the client.

a) Sometimes. I have successfully hit Teleport Home and pulled myself out of a dead TP after hitting cancel. But not always. It may depend on how quickly I reacted.
b) I don't recall. But IMs still work.
c) Not for me. I'm always safe in the destination sim after login.
d) there was one occasion where attempting to TP to the same sim caused the TP of death 3 times in a row, directly after logging in. But that was a frustrating exception, not the rule.
e) cheapie US Robotics Cable Modem Router. Model # 8004
f) I have no firewall, the Router is set to DMZ (yes I'm crazy like that)
g) It's not consistent enough for me to determine that.
h) sorry can't help you with your rash.

CPU: AMD (Unknown model) (3013 MHz)
Memory: 2046 MB
OS Version: Microsoft Windows Vista (Build 6000)
Graphics Card Vendor: NVIDIA Corporation
Graphics Card: GeForce 8500 GT/PCI/SSE2/3DNOW!
OpenGL Version: 2.1.1
LLMozLib Version: 1.1.0 (Mozilla GRE:1.8.0.12_0000000000)
Packets Lost: 931/244394 (0.4%)
Viewer Digest: ffac34f4-0452-2fcc-16b7-c2868f0923ef

http://forums.secondlife.com/showthread.php?threadid=232812

And one tech note to add that may or may not be relevant. As I've said, I've not had these crashes roughly since I started running the "CoolViewer" http://sldev.free.fr/ for a list of patches applied. But I also tried turning RIP1+2 on in the router around the same time. I've no idea how much of an impact that's having except in theory I'm getting more direct routes to the servers. TBH though, I think it's either more to do with running the patched client software, or I'm just lucky all of a sudden!

What I've noticed is that, during login, when the progress indicator pauses for a minute or so (sometimes up to 3 minutes) on 'Initializing World...', I am guaranteed that I won't be able to teleport and that my group list will be empty.

Now, I've also had the failed TP when the group list is populated, so I reckon it's a co-symptom, for want of a better phrase, rather than directly related.

MacBook Pro (series 1)
Any client.

I put the Mac into my DMZ when using SL.

2008-01-09T06:59:51Z info src=24.64.230.3 dst=86.155.143.105 ipprot=17 sport=16029 dport=1027 Unknown inbound session stopped
2008-01-09T06:59:51Z low src=24.64.230.3 dst=86.155.143.105 ipprot=17 sport=16029 dport=1028 UDP Port Scan Detected
2008-01-09T06:59:51Z info src=24.64.230.3 dst=86.155.143.105 ipprot=17 sport=16029 dport=1028 Unknown inbound session stopped
2008-01-09T07:00:00Z info src=64.129.43.2 dst=192.168.1.7 ipprot=17 (layer 4 info unknown) Unknown inbound session stopped
2008-01-09T07:00:00Z info src=64.129.40.204 dst=192.168.1.7 ipprot=17 (layer 4 info unknown) Unknown inbound session stopped
2008-01-09T07:00:13Z info src=70.42.62.21 dst=86.155.143.105 ipprot=17 sport=5062 dport=22891 Unknown inbound session stopped
2008-01-09T07:02:56Z info src=24.64.137.222 dst=86.155.143.105 ipprot=17 sport=9175 dport=1026 Unknown inbound session stopped
2008-01-09T07:02:56Z info src=24.64.137.222 dst=86.155.143.105 ipprot=17 sport=9175 dport=1027 Unknown inbound session stopped
2008-01-09T07:02:56Z low src=24.64.137.222 dst=86.155.143.105 ipprot=17 sport=9175 dport=1028 UDP Port Scan Detected
2008-01-09T07:02:56Z info src=24.64.137.222 dst=86.155.143.105 ipprot=17 sport=9175 dport=1028 Unknown inbound session stopped
2008-01-09T07:05:26Z info src=24.64.185.181 dst=86.155.143.105 ipprot=17 sport=16428 dport=1026 Unknown inbound session stopped
2008-01-09T07:05:26Z low src=24.64.185.181 dst=86.155.143.105 ipprot=17 sport=16428 dport=1027 UDP Port Scan Detected
2008-01-09T07:05:26Z info src=24.64.185.181 dst=86.155.143.105 ipprot=17 sport=16428 dport=1027 Unknown inbound session stopped
2008-01-09T07:05:26Z info src=24.64.185.181 dst=86.155.143.105 ipprot=17 sport=16428 dport=1028 Unknown inbound session stopped

Time gaps are re-logging after being kicked out, and there is a significant time waiting to log on when bar reached "initialising" after a tp has failed. The first log on of the day doesnt have this wait.

I'm no techie but hope this helps.

Meri

Incidence of this wretched problem had been declining steadily (for me) since the start of Dec, from about 95% of ALL teleports at Dec1 down to fewer than 10% by the New Year.

Woohoo, I thought. I can sort of live with that.

Since yesterday's asset server upgrade .. 90%+ of all teleport attempts cause an involuntary log-out, again. Strong correlation with groups not loading is still there (for me .. I know some people don't see that).

I've got to stress (if needed), that NOTHING has changed at my end. Same hardware, same OS, same router, same firewall configuration as this time last year. For me this problem comes and goes as the servers are messed with. Come on guys, think! What did you change yesterday that might be messing with teleports?

Simple lack of availability of asset server service? (I've given up trying to script for today as well .. scripts are neither loading nor saving reliably enough or quickly enough).

GD

Gil - Just a thought, but maybe the windows DNS cache couldn't cope with the change of server layout after the downtime? I could be way wrong here, but again, perhaps worth a try ... http://vlaurie.com/computers2/Articles/dnscache.htm for details.

I'd be interested to know if the tip at the end of that article on not storing bad DNS lookups helps anyone out.

Have a new revelation to this problem
If i log in at home, cannot tp at all...100% failure...but if i type a region in the login and start there I am able to get one tp but only one, next tp attempt freezes my screen.

Jakes explaination of the issue seems to ring true, it would explain the firewall problems I am having, and sometimes the random pattern of failures.
SL has changed something since the 18.4.3 upgrade which is when all my probs started. Which seems a long time ago now.

Heh. Sorry watchers to edit. But just to make that crystal clear.
Viewer software A and Router A .. ran fine for several weeks.
Now... With no changes to router, viewer etc...
Viewer software A and Router A .. teleport logging me out again.
During the whole period, router reset every 3 to 5 days (sometimes less).

However, I notice that it also happens on Sim crossing if the sim I'm traveling too has a lot of objects/avatars and is still rezzing on my screen. If I stop before I enter a sim and let things rez a bit and go slow before I cross then crossing is ok. If I am going fast, my av often ends up flying forever (soon outside of any sim) and then the error message occurs.

Is rezzing the environment taking a higher priority than the sim crossover or is it just that my connection has lost a packet related to the crossover?

This problem only seems to happen when my camera is fixed on my avatar rather than a stationary position somewhere. Also, sitting on a prim to make the sim crossing, prevents this problem almost entirely.

My experience of the problem is slightly different from the bug description in that the teleport screen doesn't hang. The bar slowly creeps along, but no status update messages appear at all. No "Requesting teleport...". Nothing. The bar just creeps along to about 1/3 to 1/2 the way and then DING. Disconnect. (And also viewer hang for me with Windlight. See VWR-4087). An imminent disconnect is usually apparent with in 1-4 seconds because of not seeing any status messages indicating the new region is being contacted.

I have also noticed that in general it happens mostly after I've been connected for a few hours. Up to 9 hours at a go before I teleport sometimes, and then I can usually expect the disconnect to occur. It does happen sooner in some cases, too though. Sometimes, even after just logging back in, it has been almost a 50/50 whether the teleports would work.

To Jake's questions:

a) Not sure I understand this question, especially with the RART acronym. If I understand correctly, if the teleport fails, the viewer is disconnected completely and it's not possible to resume without relogging. If you however mean that can you simply cross regions before trying to teleport, I'm yet to confirm this one.

b) No. But the viewer crashes for me. See VWR-4087

c) It generally happens each time I log in – But I stay logged in most of the time. Eventually it will happen, sooner or later.

d) Mostly over time, but I don't teleport very often when I stay logged in.

e) Linksys WRT54GL running a Linux firmware from the OpenWRT project. Usual uptime is months on end.

f) The default firewall rules policy/NAT config in the router. Some minor port forwarding for other applications. Nothing in SL's usage range.

g) Will verify this next time it happens!

h) You probably want to try something else

No one knows what happened with Jake?

He used to answer or update this "blog" every 4 or 5 days... now I didnt hear any about him long time ago...

For all... im having sucesfull teleporting every time i access (in other window) internet at the same time with teleport process... 4 or 6 times... then suddenly the next time just crash in the way ALL of you know.

If someone much more "expert" wants to try.... feedback is very welcome.... not for me... but Jake and Co.

This situation is taking a lot of time....

Land from (my land): Beerbaum
TP to: Avalon Isle (where i got banned.. don't ask me why... hihhiihih)

...i receive the message "you got baned... ablabal" so the TP failed.

2nd TP to: Parioli

...and da da! Stuck on teleport... "always".

3 in 10 teleports result in "You have been disconnected from the simluator you were in..." and a blank terrain and no other avies or scenery. Up until the failed teleport, I would be doing things normally, IMing, rezzing objects etc. I would be logged into the support portal and jira and the sl wiki (all https). Granted- https communications with secondlife.com is excruciatingly slow compared to other servers in the Internet. (hint hint)

The funny thing Jake, the simulator I was IN still thinks I am logged in. Residents around me testify I never left, except very briefly (like seconds) while my first attempt to log back in triggers the simulator to log me out, and I log back in. Not the couple of minutes it takes for me to shutdown SL and restart. (as a side note, logins to Second Life have slowed down greatly with the last rolling restart – I spend a great deal of time on the PRECACHING screen that I never really have previously).

Often times Second Life viewer doesn't know it's been logged off or disconnected from Second Life and remains normally active (until I try to move or teleport). I check my SLEEK and they are logged off - they know. (this condition is easy to reproduce). But in this teleport case, Second Life doesn't know it's still logged in... Being intermittent, I can't explain when or why it happens.

Sometimes it happens at the very first teleport after I log on

Sometimes it happens after I've been logged on an entire day.

Sometimes it happens in between. All I can say is that a one-a-day occurrence has become a several times an evening occurence in the past two weeks. My on-line habits have not changed hardly, in the past 6 months.

Kind of blows the router theory out of the window?

Also another odd thing, if I wait until I get the "you have been logged out message" and log back in, my av is in a bit a "squat" position and I can't move. I can turn but not move otherwise. The only fix in this case is to relog AGAIN. Also I have notice that if I cancel the teleport before this then close the viewer, at least I only have to relog once, rather than twice.

-DB

1) Log into SL.
2( (Disable Away when idle). Let avatar stand idle for more than a couple of hours.
3) Attempt to teleport to...Cyclops
4) Receive disconnected message
5) Open a second copy of SL (before shutting down the first, but after receiving message) -
6) Log in as an alt and teleport correctly

If it were routing tables, then the 2nd instance of the browser wouldn't have experienced proper teleportation either? Something like that would require an OS restart / not an SL restart...

=====================

Not so reliably
1) log into SL
2) teleport immediately
3) fail immediately

No rhyme or reason to the immediate failures. Also, when it happens, some of the mentor and help channels have announced that a rash of it is going on - although there are many residents who are still able to transport freely at that precise moment.

Has seemed to me for a long time now that we have one failure mode but several triggers .. so possibly several bugs. These include (at least)

i) The time-in-sim related version documented by lum & Fluf and others

ii) The server incompetency version that afflicts me (correlated with other failures such as no groups loaded etc)

iii) Particular region related versions?

iv) heavily loaded SL versions?

.....

So in a sense the real bug is that the viewer doesn't handle a failure to tp correctly .. the bug is in the error handler .. or possibly it's the sim-to-sim protocol for handing off an agent that needs a really critical eye cast over it .. it seems to be broken almost by definition since it has an 'indeterminate' state where the two ends can't agree on what happened.

1) Don't log in to "Home" or "Last Place", type a region name to log in, this I find will give me at least 3 successful tp's before logging me out. (Oh yeh forgot to say, dont stay in region too long or it won't work)

2) To get more successful tp's I find that clicking constantly on my emails "send and receive button" while tp bar is moving, helps too! ( Mad Hey?)

3) Havn't tried standing on my head whilst drinking coffee yet, but if anyone has, please let me know if it works

What a sad life we lead!

Pick two sims that are not close together.
One will be your starting sim and the other will be your destination.
Log in and go to your starting sim.
Add a packet filter blocking all ip traffic between you and the destination sim.
Teleport to the destination sim.

Expected result: the teleport times out and you are left where you were in the starting sim.

Actual result: the teleport times out and you are logged out of secondlife.

Firstly, anecdotal evidence in SL tells me this is happening to a lot of people. They just fail a TP, sigh and relog after they have been at a club or gig for a while and say "ah well, it's SL, what do you expect?" (or words to that effect).

Second, there may be something of a presence clue. Now I've witnessed a few people fail to teleport away from me, if they get as far as vanishing from my view, reasonably often I don't get a "x has logged out" message, and IM lets me chat to them is if they are on-line. It's only when they re-log I briefly get the IM "will be saved" message usually for one line before SL then tells me they are on-line again almost immediately. Don't know if that helps you track down the point at which they get dumped out, but thought it might.

Using the AT&T routers causes above described TP failures every single time.

Using the Valley Internet router allows me to play Second Life like in the old days.... sigh reminiscing....

Anyway, I have been playing SL from home using the same AT&T DSL router since Oct 2006. So there is something that has changed in the latest releases that seems to be incompatible with the AT&T DSL router/firewall setup.

Again, this is something I have noticed in my case, and may not be true for everyone. I have checked this on several different days over the past couple months, and the same results appear each and every time.

I can only play SL now when I am at my friend's house who uses Valley Internet.

Besides, it has uses in a covert sort of way. If I know someone has been on-line for ages without a break, I can now pretty much assume that they have only been in the one sim the whole time.

Suffering heavily from it, over the last couple of days I noticed a 100% success rate when teleporting, irrespectively of time spent in a region ( from nearly none to up to several hours).
IF I teleported from the map after it had loaded completely...

Inconclusive test with an alt to teleport as simultaneously as humanely possible from the same PC via the same router show that it can happen, but only to a very small percentage, that both teleports fail so spectacularly. Normally at least one of my mes gets through...

I've come to hate that bug from the depth of my little heart...

a) assert server is under load (many users online (> 50.000))
b) when I stay more than 1 hour in a place

The TP attempt takes a long time and when I click 'cancel' during the process I find myself in nowhere with no prims around me and also the "continue/quit" dialog.

Fatal effect which was also reported here several times: After logging back in I am unable to walk, fly and teleport.
I have to log out and log back in a second time for everything to work 'normal' again.

About 15% of my teleport attempts are affected by it and it happens only with viewers after 1.18.4 (windows version)
RC 1.19 is also affected. No changes in my network config or router for about 3/4 year

athlon62 X2 4400+, 2 GB, geforce 6800 GT
windows XP pro

I'm a builder, the last time I used TP while I was building, I lost all last modifications I did before TP...
Regularly, when I relog, I'm freezed and can't do anything else than chating ... (telling friends I re-relog, cause I can't move)

This problem has been reported FOUR MONTHS ago, so PLEASE ask your developpers to check this out SOON for real, this time....

Enjoy SL, be patient...

Click Continue to look at existing IM and chat.
You will not be able to perform any other operations.
Click Quit to exit Second Life immediately.
2008-02-11T21:41:44Z INFO: LLMessageSystem::disableCircuit for 64.129.41.86:13006
2008-02-11T21:41:44Z WARNING: disableCircuit: Couldn't find circuit code for 64.129.41.86:13006

I have to say that I have mixed emotions now... I was having problems with TPing my avatar in my iMac PowerPC since a couple of months ago... i posted several comments trying to give more information regarding this issue. Well, i have to say now, that i got a new laptop not a MAC but a PC with nVidia Quadro Video Card, etc. and i downloaded the 1.18.5(3) version of SL and start to doing some tests.... well as far as today (one week later) all my TP's were been sucesfull. An interesting point here... my router is the same... nothing change, no config., no new firmware, not even a reboot... I just changed the Computer....
I will keep doing test and trying to post feedback to this forum in order to find out where is the problem.... because I rather play with the iMAC than the PC... I saw some comments regarding that in some cases people change the computer and the issue is still there... well i dont know where's the catch... but this wasnt my case...

...unbelivable...

Jake:If you need more info about the config of my laptop, let me know