Issues with disconnects

[akleshchev]

Environment

Second Life Test 7.2.0

Description

Some 'disconnect' issues that don't make much sense came up. No solid solution yet, but some improvements can be done:

• Attemp to send logout when disconecting even if we lost network (should speed up relog in case it gets through)
• Either revising the 45s delay or extending timer if some kind of communitation/changes do go on not applicable, the timer in question was not responsible for the disconect
• Sanity check bandwidth input.
• Improve logging/user feedback for reasons of disconnect

Reproduction steps

Kady's and Strawberry's logs in slack

Strawberry teleported over couple locations before going back into initial region which failed.

Synthetic testing:

• Use 'force disconect' from develop menu and check that logs have a logout attempt
• Attempt to set unreasonably high or low ThrottleBandwidthKBPS in debug settings (like 20MB/s) then check value in 'about'

---

This repo is using Opire - what does it mean? 👇

💵 Everyone can add rewards for this issue commenting /reward 100 (replace 100 with the amount).
🕵️‍♂️ If someone starts working on this issue to earn the rewards, they can comment /try to let everyone know!
🙌 And when they open the PR, they can comment /claim #4570 either in the PR description or in a PR's comment.

🪙 Also, everyone can tip any user commenting /tip 20 @akleshchev (replace 20 with the amount, and @akleshchev with the user to tip).

📖 If you want to learn more, check out our documentation.

[kylelinden]

Earmarking for 2025.07 to sort out what's going on with the disconnects.

[Ansariel]

I can only reply what I mentioned numerous times and 10+ years before already: Teleports are missing a failsafe mechanism. To recap what happens during teleport - again:

• Viewer receives circuit URL for the destination region
• Once the viewer signals it received the URL, the source region terminates the circuit with the viewer
• Viewer tries to create circuit with the destination region
• Insert random event that cause attempting to create a circuit to fail
• Viewer has no circuit with the destination region yet nor with the source region anymore, nor does it attempt to create a circuit again
• Wait 60 seconds for "You have been disconnected from the region you were in" message

Can you spot the problem?

[akleshchev]

I'm aware of that particular set of issues, but not much I can do there, will see if we have any tickets to adress it.

[akleshchev]

@Ansariel created #4589

Viewer has no circuit with the destination region

In case that caused this ticket, teleport almost finished, but server was taking a lot of time to processes attachments so viewer timed out.

[Ansariel]

I'm aware of that particular set of issues, but not much I can do there, will see if we have any tickets to adress it.

Well... one might ask the question how it is possible nobody came up with a solution for this issue that has been known for over 10 years already? This is a common thing that happens a lot, especially if heavy crowded regions are involved. It's not just that you have to reconnect, but changes to attachments also get lost (unless you were smart enough to detach them temporarily after making changes)! Is it such a difficult task to have the source region wait with terminating the circuit to the viewer until the destination region confirms a new circuit has been created?