Problem :
When CloudLinux applies a limit it gives an error message and their Apache modhostinglimits shows a 508 Server Busy error message. Litespeed doesn't detect this error message and shows a 503 internal server error instead.
When it happens :
This happens in two cases :
1) If a user goes over his max Entry Process limit
2) If a user's account uses up too much CPU (the CPU queue gets too long)
Why it happens :
Currently Litespeed isn't compatible with these two situations. It doesn't detect Cloudlinux's error, it tries multiple times to connect to PHP and shows a 503 internal error instead of 508 server busy error. 503 errors make Litespeed to restart.
Litespeed's current workaround
You can fix the first problem by increasing CloudLinux's Entry process limit to be larger than Litespeed's limit but this means you can no longer set Entry process limits per account or per package.
When Litespeed's workaround doesn't work :
You can't fix the second problem as if you remove CloudLinux's CPU limits that removes the whole point of CloudLinux.
What could happen
If multiple accounts hit their max CPU queue limit and both start trying to show 508 errors which are not recognised and 503 errors are show instead, litespeed with start a series of restarts. If the server doesn't have much ressources left restarting litespeed will not help and could cause unavailabilities or make the server crash.
My Feature request :
My feature request is for Litespeed to detect CloudLinux error messages and show a 508 error message instead of a 503, thus not restarting Litespeed when a CloudLinux limit is hit.
Who need's this Feature ?
I'm pretty sure a large percentage of your customers use both CloudLinux and Litespeed so this request should be put on a quite high priority. Any server using both Litespeed and CloudLinux need's Litespeed to be able to detect CloudLinux Errors.
When CloudLinux applies a limit it gives an error message and their Apache modhostinglimits shows a 508 Server Busy error message. Litespeed doesn't detect this error message and shows a 503 internal server error instead.
When it happens :
This happens in two cases :
1) If a user goes over his max Entry Process limit
2) If a user's account uses up too much CPU (the CPU queue gets too long)
Why it happens :
Currently Litespeed isn't compatible with these two situations. It doesn't detect Cloudlinux's error, it tries multiple times to connect to PHP and shows a 503 internal error instead of 508 server busy error. 503 errors make Litespeed to restart.
Litespeed's current workaround
You can fix the first problem by increasing CloudLinux's Entry process limit to be larger than Litespeed's limit but this means you can no longer set Entry process limits per account or per package.
When Litespeed's workaround doesn't work :
You can't fix the second problem as if you remove CloudLinux's CPU limits that removes the whole point of CloudLinux.
What could happen
If multiple accounts hit their max CPU queue limit and both start trying to show 508 errors which are not recognised and 503 errors are show instead, litespeed with start a series of restarts. If the server doesn't have much ressources left restarting litespeed will not help and could cause unavailabilities or make the server crash.
My Feature request :
My feature request is for Litespeed to detect CloudLinux error messages and show a 508 error message instead of a 503, thus not restarting Litespeed when a CloudLinux limit is hit.
Who need's this Feature ?
I'm pretty sure a large percentage of your customers use both CloudLinux and Litespeed so this request should be put on a quite high priority. Any server using both Litespeed and CloudLinux need's Litespeed to be able to detect CloudLinux Errors.
