Recently, there have been frequent disconnections and reconnections when using the editor (averaging once every 10 minutes), during which changes are often lost.
The experience has enabled draft mode, but even if it is submitted, once disconnected and reconnected, the script changes will also be lost
But if you look through the version history, you can see that the script should have been submitted, but for some reason it was not read correctly
Moreover, frequent disconnections, reconnections, and loss of changes have seriously affected the development process of the game.
(All people accessing this scene will experience disconnection and reconnection)
From reviewing the report, it looks like there are two issues going on:
Frequent disconnections
Drafts not committing
On the first issue, I’m curious what is happening at the time of disconnect. Are users trying to save the game? During the script commit? Do all users get kicked out?
On the second issue, I’m curious if this is still happening. We have rolled back a recent change to Studio that caused some issues for Drafts. Could you check your Script Version History and see if the message “Current Source does not match latest commit.” is still there?
Hi, thanks for reply.
About the first issue, I’m not sure, I didn’t actively save it, but it may have been a auto save. But after losing the changes, I often proactively save them in an attempt to avoid loss (Despite this, save failed still occur frequently).
Yes, everyone connected will disconnect.
About the second issue, I need to observe because after trying to replace the proxy last night (1:00 UTC+8), there were no more disconnections, reconnections, or draft losses (originally using Southeast Asian nodes, later changed to American nodes)
Some disconnections may be happening due to network conditions. Are you using a VPN to connect to the internet? If possible we recommend turning VPN off to improve save success rate. If not possible, a VPN connection in the US would be preferred.
I think my network is fine, and I can only access Roblox using VPN.
Of course, the previous disconnection issue no longer occurred after switching to a node in the US region, but my colleague is using a node in Hong Kong and has not experienced any disconnection recently.
Oh, unfortunately, the issue of disconnection seems to have reappeared. My colleague is using a node from the United States.
I am editing other experience and have not encountered any disconnection issues yet.
I think it is not a network problem, as it has been before, only a few experience have frequent disconnections, not all.
At present, the situation of disconnection mainly occurs in experience with a large amount of resources, and has not yet been encountered in experience with limited resources.
Moreover, it is only in nodes outside the United States that disconnection occurs (due to regional issues, we can only access Roblox through VPN)
It seems that the region of the editing server depends on the location of the first connected user?
Today, due to the first connection using a node from a non US region, the disconnection occurred again, and the script submitted through the draft still experienced a loss issue
Hi, thanks for getting back to me with the logs! I took a look at both the client and server logs.
The server started in US/West. To confirm, the server will start in a location according to the first client that initiates the session.
From the client, I see the disconnect is due to error 277 (connection lost). Around the same time, I see HTTP timeout errors from various URLs which suggests the client has lost internet connection.
From the server, I see multiple users disconnect with the same error 277 (connection lost). Are multiple users connecting from the same location?
I see that the place save succeeds from the server so it doesn’t seem to be an issue specific to this place. In the past we’ve had RCC shutdowns due to memory overuse, but I don’t see that here. There doesn’t seem to be data loss in this case.
So, if the user loses internet connection during the commit, the request won’t reach the RCC server. Once re-connected, can you re-try the commit?
Also, let us know if you see “Current Source does not match latest commit.” again
Thanks for check, currently my colleagues and I have changed all VPN nodes to the US region due to previous issues, and the problem of disconnection no longer occurs
I will find time to try replacing the node back to the Asia Pacific region and try again
Supplementary explanation
With regard to the client’s loss of Internet connection, I think my network should be OK, because I was opening multiple experiences at the same time, only the experiences with too much content were disconnected and reconnected, and the experiences with too little content (not empty, under development, without too much resource content) were not disconnected
We are connected in the same location, but I think it depends on the VPN node, which I am not sure about.
In fact, I am quite puzzled about the previous disconnection issue. If the script is successfully commited, does it mean that the changes in the experience should also be saved? But at that time, changes made in the experience after disconnection were usually lost, and manual saving would also fail due to network issues.
Per-chance does your VPN provider rotate your IP? I’m not exactly sure about Team Create but (afaik, unless the behaviour has since changed) the Roblox client cannot handle IP address changes and will kick you for your internet connection if it changes; it may be the same for Team Create.
Probably not, although everyone disconnects at the same time, as I said, disconnections only occur in experiences with a large amount of content
And VPN has multiple nodes, so the probability of using the same node is not high. Some people use other VPN providers, but everyone disconnects at the same time
To answer your question, it is possible that the Script Commit succeeds but the Place Save fails. Place Save failures can happen for various reasons, but are most common with large places. For places larger than 20MB we upload in multiple chunks. If this takes a long time, the TeamCreate server may experience lag (and possible timeouts). We’re making this more asynchronous in an upcoming fix.
