From previous news you know we did not get a new package rolled to the main channel. Thursdays Server Beta meeting gives a bit more information on what to expect for next week. For now Maestro Linden thought the Interest List package running on Magnum was the best bet for getting promoted next week (5).
Magnum is not completely trouble free. Some people are reporting problems with Bots…. aaaw. Sorry. Those running bots are people too and are probably paying the Lab ore than most Second Life™ users.
It seems LibOpenMetaverse bots go into some type of high bandwidth use, pulling 100kb/sec. Restarts of the bot and region don’t seem to help.
You can see the difference in the graph. The day of the rollout 1/23, you can see the jump in data transfer. Below you can see the rate increasing during the first four minutes of a bot relog.
Obviously that is a problem. But, is it a problem with the server or with the bot? At first I wasn’t sure and I doubt anyone was Thursday. I thought that because the Morris region where the meeting was held is running the same software Second Life RC Magnum 13.01.17.269162 (the version on Morris may have changed now). None of the people there were seeing high bandwidth use. That may only indicate the ADITI region is immune. The higher use AGNI region may be running into problems.
Andrew Linden looked at the problem. He is changing the initial camera properties server side to more sane defaults and will try to get this out in an update. That has to go through QA, so it may not make it in time for next week’s rollout.
Andrew also looked at the libOpenMV’s code. It appears that the Test Client was being run and it supports sending AgentUpdate messages. This would account for the high bandwidth use. Maestro said, “It is not clear whether a valid camera ‘Far’ clip is sent in your case, or if it is sending zero. The values that are put into the AgentUpdate message depend on some config files. It makes sense to me that someone running a chat bot, or a parcel search bot, would set the camera’s Far clip to zero to minimize streaming of visual objects which the bot would probably just ignore.”
So, it is a combination of how the bots are accessing SL and triggering the problem and how the server is responding with default camera values. After reviewing the problem Andrew has more confidence in his server side fix. We just have to hope it works and makes it through QA in time for next week.
If you run bots and are seeing this problem can read the information in the Deploys thread.
Unfortunately problems are not just limited to the bots. Others are reporting:
- Chat duplication
- Attachments displayed as worn in inventory not appearing at all
- Attachments not appearing completely, i.e., some parts of worn link-sets are not visible. No pattern detected.
- Some objects in world not appearing at all or incompletely
- Avatar not always able to enter region “in a timely manner” failures common
- Upon entering region, ones own avatar appears frozen in place but is probably actually moving when controls are operated (see next item)
- After entering region by teleport, my avatar hovered at the teleport destination for many seconds before suddenly “warping” in the direction I was trying to travel by walking. This is not the normal “object interpolation” jump seen when traveling linearly and receiving sparse course updates.
- Outfit changes are MUCH slower in the affected region than in the baseline region
- I can’t move on Magnum Sim my avatar seems to be frozen, but it actually moves but i don’t see it
- The world is loading very slow
- Other Avatars don’t load @ all
- High Paket lose all the time : Paketverlust: 10.156/69.062 (14,7%)
Whirly Fizzle, Firestorm Developer, reported that Firestorm Support has been hearing about : “…high packetloss/high ping sim on Magnums only and when an affected avi arrives on a Magnum, they can see other avatars fine UNTIL they move their avatar or move cam, then all avatars derender. It’s rather strange. Seems to only affect a minority but they all say pretty much the same thing.”
This may only be a problem for those with a slow connection. So, if you are seeing the problem, look through Troubleshoot Your #SL Connection for how to test our connection to the Lab’s servers. Report the results and describe your problem in a JIRA report and the forum thread. It works best if you file the JIRA and list it in your forum post.
There was a BUG filed for Magnum that also mentioned some tennants not being about to see their houses.
So, I’m not at all sure the Interest List package on Morris will rollout next week. By the end of the meeting Maestro was thinking the Region-Crossing package had a better chance of rolling out.
If you are seeing similar problems, get your information into the JIRA and the Deploys thread. The Lindens need to know if only you are not seeing avatars or if others are seeing the same avatars disappearing. So, ask around when you see this problem.
Simon Linden says, “We should have last week’s maintenance RC back next week too.” That is the fix for places falling out of search issue.
ADITI Problem Continues
We have not heard anything about whether the Lindens have added more disk capacity to ADITI’s back end or not. Some people are still running into problems with inventory and rezzing things when they change their password.
For now it is a trial and error process. They were manually clearing disk space. If that is still the case, then it is a matter of changing your password shortly after some disk space is cleared up.
Remember. Lots of people change passwords regularly as a safety precaution. I suspect some large percentage of those do not use ADITI. The Lindens are figuring out how to cull or handle that data load so actual ADITI users can have working inventories.