However, since the bad packet handling under Linux and Windows appears to be completely different, I though a new thread might be a good idea.
My experiences:
On a Bad Packet day, where even repeated relogging does not improve things much, I have seen the following:
Linux gets a Bad Packet value of 150%% and higher. I have seen 250%% at one time. Is the %% a sign for per 1000 (pro mille)? Or is it percent with one too many % sign? If the bad packet rate is arounf 150% initially, it usually settles down to 0% when the initial download is done. Since Linux SL throws the cache away, the whole scene has to be fetched from the servers, so a LOT of download volume. During this time,I have seen bandwidth usage of 1500 Kbps, tho SL is set to 1000 kbps max. In the log window, it can be seen that the bandwidth gets reduced (by the SL client?) stepwise, 100 kbps steps, until the bad packet rate drops; then it immediately starts climbing again, in the same 100 kpbs steps. The textures all load into the fuzzy warmth I know from SL Linux, and sharpen when I mouse over, or right click. Only mouse over works for avs and their clothing etc, and trees and terrain textures are a matter of luck.
The Windows client under Windows (not wine) does not get these bad packet levels. It gets like 30 to 50 %%, and quickly settles down to 0 %%. However, the bandwidth sticks at like 30 kbps, and some textures just stay grey forever (it seems like). The bandwidth does not seem to climb again, to get the rest of the textures, it seems to max out at 50 kbps or so. Right clicking and mouse over work, eventually, but at the rate you would expect when the bandwidth is pegged at 30 kbps.
Edit ---
I had something on how wine behaves here, but my memory is too vague, so I removed it.
--- End Edit
Running winmtr.exe via wine does not show any major problems, other than that some router in the States (I am in Europe), before the LL servers, has a "few bad packets". I haver a cheapo provider
Traceroute on data.agni.lindelab.com gets blocked at the entrance to the LL server farm
. It is still ok to check up to that point. Funny that winmtr.exe goes through; is there a special port that should be used with traceroute?Anyone else had similar experiences?
The Linux Client appears to be getting a "bad packet" reputation, but when I look, its because the Linux Client always tries to maximise the bandwidth dynamically, second by second, and the Windows client just drops the bandwidth and leaves it there.