thecode

VM logs should be fixed with Docker Folder 2022.06.19 The VM folder expand is still broken

So I added a fake cookie (named it "test"), deleted the ca_data and started increasing its payload until the VNC session no longer connects. When total of all cookies is 3159 bytes it still works, increasing the cookie by one byte to 3160 breaks VNC and reducing it again by 1 byte fixes VNC. So the limit at least on my system is 3159, now need to figure out where does this limit comes from EDIT: also checked without SSL, the limit that still work without SSL is 3522

It is not the cookie itself, there is no problem with the size of it my ca_data cookie is 438 bytes while the rxd-init which I first suggested removing is 541 bytes. The total size of all cookies on my system right now which already breaks VNC is 2816 bytes so surely all of them does not even reach the limit for a single cookie. However the combination of the size of the cookies breaks something. I will try to find the limit by editing a cookie and reducing the size until it starts working. @Squid just to be clear this is not CA related. There are also other big cookies. I showed this as an example since it is easy to delete the CA data and let CA create it again, while for example deleting the rxd-init is not easy to reproduce back the problem since it starts small and get larger with time.

The only problem is that it requires doing it every few days. I have also identified that it is not specifically the `rxd-init` but probably a combination of big cookies. I know now how to reproduce it every time without clearing cache, but still not how to solve it. For example community apps also adds a big cookie to store it's settings so deleting it (ca_data) also fix the problem, going to CA settings brings back the cookie which will again break VNC. The error even hints that something happens during read: recv() failed (104: Connection reset by peer) while reading upstream So I guess it would be possible to increase the read buffer somewhere, but this would have to be looked by someone from @limetech

Can you explain what doesn't work? did you remove the old one, searched and added Docker folder again?

I didn't notice it until you mentioned it, I can confirm it is broken and while testing it I found out that the logs icon is also broken. Trying to view logs for a VM will open a window with the following error: `Error: No such container: ....`

Yes I did, see my explanation at I linked this so we make 1 discussion instead of two. This one should be closed IMO. Clearing browser cache solve this for a period of time (I can't spot the exact period, hours to days), I did identify that the `txd-init` or `rxd-init` cookies are causing the issue, when you clear cache you also clear these cookies.

This is already reported at

I tried to find out what is stored in the cache that causing this problem, so instead of clearing all application cache I looked at the stored data. It looks like the `rxd-init` and `txd-init` cookies are too long and break something. I would appreciate if someone confirm my findings (as it takes hours to days for this to happen) - For chrome users: - When the problem is active and NoVNC does not connect - From the VNC window with the error message press F12 - Select Application and cookies on the DevTools window - Delete the `rxd-init` cookie - Press Connect on the NoVNC window

I didn't notice you wrote about something else, I can confirm the toggle behaviour.

Follow This post is also pinned at the top of the page

Yes

That was mentioned by @Squid it is currently disabled by purpose since it does not work correctly and is kind of a hack of the dashboard system (I tend to agree with that)

@Groto it doesn't matter that it is a fork, community apps repo was updated to include this, will show only from unraid 6.10.0 and above.