We've significantly improved websocket handling on our agent. These websockets are used to keep a live connection between our servers and the agent, facilitating real-time communication. The new version is more responsive, reducing our time to detect a device going offline from three minutes down to just thirty seconds.
We can now detect and resolve common yum errors that can occur during patching. Once the error is resolved we automatically retry the patch install.
You can now use the "Original Size" scaling option even if you don't have enough room to fit the entire desktop in your browser. This will encode the desktop at its full size and then downscale it in the browser. This render-side downscaling can still result in superior quality compared to auto-scaling.
We've added to tooltips to any disabled scaling options, explaining why they're disabled. Typically this is because there isn't enough room in the browser for a larger scale.
The scripts page has a new header that provides a link to our community scripts repository.
We've improved the retry logic on our remote control page. This should allow us to recover more failed connections automatically and provide a button to manually retry if we can't.
The unit of time displayed for a "run script" monitor in the monitor's table now matches the unit selected during monitor configuration.
We've made improvements to our "run a process as the current user" code. For now, this will aid in troubleshooting remote control problems as our encoder will be giving us richer logs. It also paves the way for other user process improvements such as running a script as the current user. More to come!
Bug fixes
Some Windows Servers were missing a security score because of a parsing error on the agent. This has been resolved and those scores should automatically update.
Fixed a bug that caused some groups to be inaccessible when using the group search.
The sort order of groups in the sidebar is now case-insensitive.
Shrinking the remote control sidebar to its minimum width will no longer cause the stream to be displayed off-center.
Our systemd configuration specified that we wait two minutes before restarting the Level service. This has been reduced to five seconds.
We've increased the timeout for disk monitor readings to make sure they can be collected on slower machines.
Fixed a bug on the Windows agent that caused Level to restart after every terminal session.
A device monitor error is now kept up-to-date while it's open. Previously we were sticking with the first reported error message.
Fixed a bug where the dark theme from the remote control window could get applied to the rest of the application.
The UI will now properly update device search results if a change in search term causes the same devices to come back in a different order.
Fixed a bug that prevented remote control from working on scaled displays.
Uninstalling a monitored service on a Linux device could sometimes cause the monitor to get into an alert -> resolved -> alert loop. This has been fixed.
