Cool VL Viewer forum

I've been noticing periodic "stalls" where the viewer comes to a halt in the middle of something.

I had the debug console running, so I could see if there was a message that came up at the same time, and it's one of these:

INFO: LLAppViewer::idle: Kills on unknown objects: ###
(Where ### is the number of objects it's apparently killing)

In the mall I was in tonight ("MakersMart", formerly "Sculpts Only"), it seemed tonight that the greater the number of objects being killed, the longer the "stall" happened for. Back in the home sim (where there's substantially less scripts and objects) it doesn't seem to happen as strongly.

I don't see a matching set of message before each occasion of the "LLAppViewer" message, though.

If I wasn't able to provide enough details, let me know and I'll attach a log.

The object kills are normal (they are derezzed objects, i.e. objects not in the FOV any more) and don't cause any hiccup.

I cannot diagnose your problem without the full logs (please, read and respect the rules of this forum), but my guess is these were just the normal hiccups you experience a couple minutes after a TP, when the viewer disconnects from the sims surrounding the place you TPed from. There is no way to avoid those.

Logs are attached of this round of stalls - I wasn't in the same location. Plus an added bonus of screenshots, wehre I had the debug console running and the big map up so you can see that the sim is actually EMPTY - there were no teleports in OR out during this timeframe. (I'm used to teleports in/out causing stalls - that wasn't the case for the last post, and isn't the case now.)

https://lh5.googleusercontent.com/-_cin ... 252001.png
https://lh5.googleusercontent.com/-C6OG ... 252002.png
https://lh3.googleusercontent.com/-OiyB ... 252003.png

I was not speaking about incoming teleports from other avatars, but from the fact that after *you* teleport into another sim, approximately 2 minutes after your arrival, the viewer gets disconnected from the sims that it was connected with before your TP, and this does cause a few unavoidable "hiccups"...

Anyway, thanks to the log, I can see that your problem is not related to this, neither with the object kills, but rather to an issue with the network. The relevant lines are the numerous:
which denote a failure from the server to reply to the viewer and affecting HTTP texture fetches.

Try the following, and see if it cures your issue:

• In Advanced -> Network, disable "Multi-Threaded Curl" and restart the viewer.
• If the above fails, try lowering the "Max HTTP requests" setting in the "Network & Web" tab of the Preferences floater.
• If the above fails, try disabling the "HTTP fetch textures" setting in the "Network & Web" tab of the Preferences floater.

Normally, the 503 HTTP error denotes a server-side issue (non-responding server), however there could be some other reason related to your own network, such as a bad DNS server (too slow, or unreliable: try changing for another in your network configuration), or your firewall blocking the HTTP requests from the viewer to the texture server.

In any case, I cannot reproduce this issue here (and I did visit the Arcania region, corresponding to your log file)...

Ah, I don't notice that first couple minutes so much as that is usually when I just stay put while everything's rezzing in anyway.



Did that, annnnd, here's the resulting image. It did seem to reduce the random stalls, only now textures are taking forEVER to load - if they ever do. The included picture is after close to 10 minutes hanging around by a lucky chair, moving the cam, clicking vendor boards, and generally trying all the "old skool" methods I can think of to try and kickstart image loading.

https://picasaweb.google.com/lh/photo/t ... bedwebsite
(you can use the magnify function there if you want to get a closer look at the interface floaters)

I realize you've said the problem is likely at my end; I'm just sharing in case something else becomes evident this time that I'm missing. I did notice more of those "curl get failed" messages, even though I have curl turned off per your steps.

I'm hesitant at this time to lower the "http requests" or turn it off completely because textures are now already taking an eternity to load with just multi-threaded off. If nothing new comes up after you look at this, I'll likely turn curl back on and just bear with the stalls.

(I'll attempt to attach the log in a separate message. Having some difficulty this morning.)

Mrrrrgh. Ok, I give up trying to attach a file today. Is there an email I can send it to?

Reducing the number of simultaneous HTTP requests is something to try, since it makes no sense to make plenty of requests that end up stalling because of network issues. Try 16, then 8, and see what happens.

Disabling the HTTP texture fetches will simply revert to v1.23's way of doing things (pulling textures from the sim instead), so while normally slower it might be best for you till you fix your network issues.