Brave Browser uses 100% CPU when back end servers are down
1–7 of 7 posts
Re: Brave Browser uses 100% CPU when back end servers are down
#2The problem fixed itself when backend servers were reachable again.
Re: Brave Browser uses 100% CPU when back end servers are down
#3The problem fixed itself when backend servers were reachable again.
I don't know what the design decision is here, but in a sane browser, there should be 0% reliance on "backend servers" in order for it to be usable.
Re: Brave Browser uses 100% CPU when back end servers are down
#4The problem fixed itself when backend servers were reachable again.
I don't know what the design decision is here, but in a sane browser, there should be 0% reliance on "backend servers" in order for it to be usable.
I assume it's just out-of-control polling. The fix is probably to add a stepping timeout on errors. Not ideal, but not a very egregious oversight given that a browser will nearly always be used in an environment that's at least internet-connected.
Re: Brave Browser uses 100% CPU when back end servers are down
#5tweet at the founder/CEO, he appears to have genuine concern over the project and is quite engaged on twitter
Re: Brave Browser uses 100% CPU when back end servers are down
#6Earlier quoted context omitted.
I don't know what the design decision is here, but in a sane browser, there should be 0% reliance on "backend servers" in order for it to be usable.
I assume it's just out-of-control polling. The fix is probably to add a stepping timeout on errors. Not ideal, but not a very egregious oversight given that a browser will nearly always be used in an environment that's at least internet-connected.
Lack of internet connectivity should be a basic scenario any browser should have a plan for.
Re: Brave Browser uses 100% CPU when back end servers are down
#7Earlier quoted context omitted.
I don't know what the design decision is here, but in a sane browser, there should be 0% reliance on "backend servers" in order for it to be usable.
I assume it's just out-of-control polling. The fix is probably to add a stepping timeout on errors. Not ideal, but not a very egregious oversight given that a browser will nearly always be used in an environment that's at least internet-connected.
Whoosh!
Why is a browser polling backend servers in the first place? I thought it was to fetch urls upon end-user request?