----- Original Message ----- > Most likely the event loop timeout is too short for the VM. The > shortest timeout should be at least 10 msec in most cases to avoid > busy waiting. Practically speaking, we should be able to use a 1 > second timeout and just interrupt the thread via the eventfd > mechanism.
> Can we get a specific repo list: Virtual Box Version + OS type and > version + ATS version and build flags (if any). We have a bug for that: https://issues.apache.org/jira/browse/TS-1365 -- i > Thanx, > john > On Tue, Mar 19, 2013 at 3:28 AM, Matthieu BIENVENÜE < > [email protected] > wrote: > > I've tested on a server with a Quad-Core AMD Opteron(tm) Processor > > 2376 HE (with on core setup for the VM with ATS) and I've tested on > > my own laptop with an Intel Core 2 Duo P7450. > > > Le 19/03/2013 11:24, Igor Galić a écrit : > > > > ----- Original Message ----- > > > > > > > Am 19.03.2013 09:33, schrieb Igor Galić: > > > > > > > > > > > ----- Original Message ----- > > > > > > > > > > > > > > > > Hello ! > > > > > > > > > > > > > > > > > > > > > I've just installed ATS on a Debian 6.0 (but with squeeze > > > > > > > > > > > > > > > > > > > > > repository > > > > > > > > > > > > > > > > > > > > > in > > > > > > > > > > > > > > > > > > > > > order to install ATS using packages) on Xen server host > > > > > > (other > > > > > > VM > > > > > > > > > > > > > > > > > > > > > on > > > > > > > > > > > > > > > > > > > > > the > > > > > > > > > > > > > > > > > > > > > same host work like a charm). > > > > > > > > > > > > > > > > > > > > > ATS is now setup and works as a reverse proxy but it > > > > > > consumes > > > > > > CPU > > > > > > > > > > > > > > > > > > > > > even > > > > > > > > > > > > > > > > > > > > > with no connection (that's a lot for doing... nothing !) : > > > > > > > > > > > > > > > > > > > > > For example the result of the top command : > > > > > > > > > > > > > > > > > > > > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ > > > > > > > > > > > > > > > > > > > > > COMMAND > > > > > > > > > > > > > > > > > > > > > 1929 traffics 20 0 93192 31m 1964 S 11.9 53.7 0:02.45 > > > > > > > > > > > > > > > > > > > > > [ET_NET > > > > > > > > > > > > > > > > > > > > > 0] > > > > > > > > > > > > > > > > > > > > > (don't take care of the memory the machine only have 64MB > > > > > > of > > > > > > > > > > > > > > > > > > > > > memory, > > > > > > > > > > > > > > > > > > > > > for > > > > > > > > > > > > > > > > > > > > > testing only). > > > > > > > > > > > > > > > > > > > > 64 MiB seems like really little. > > > > > > > > > > > > > > > Did you disable caching? Otherwise I cannot imagine ATS even > > > > > > > > > > > > > > > starting > > > > > > > > > > > > > > > properly with that little RAM > > > > > > > > > > > > > > but this has nothing to do with the CPU usage > > > > > > > > > > i have here a dedicated trafficserver-vm in production > > > > > > > > > > with 4 GB RAM and 10 GB RAW-cache-disk which also consumes > > > > > > > > > > all day long some percent CPU usage while any other server > > > > > > > > > > is completly idle from view of CPU load at night > > > > > > > > > ACK, ACK, ACK. > > > > > > I was just surprised it started! > > > > > > I have noticed this behaviour a long time agoe on FreeBSD, I > > > > > > didn't know it's spread over to Linux in such a way already. > > > > > > How many cpus/cores/threads does your server have? > > > > > > @devs: Has anyone observed this kind of behaviour on a system > > > > > > where there is a usable DTrace? > > > > > > -- i > > > > > > Igor Galić > > > > > > Tel: +43 (0) 664 886 22 883 > > > > > > Mail: [email protected] > > > > > > URL: http://brainsware.org/ > > > > > > GPG: 6880 4155 74BD FD7C B515 2EA5 4B1D 9E08 A097 C9AE > > > -- Igor Galić Tel: +43 (0) 664 886 22 883 Mail: [email protected] URL: http://brainsware.org/ GPG: 6880 4155 74BD FD7C B515 2EA5 4B1D 9E08 A097 C9AE
