hi,
i am currently using for a sport event 6 web nodes running lightspeed standard version 2.1.14 behind load balancers, serving around 800 requests/sec per node, writing 200/mb logfiles per node. php is also involed in using lsphp withj eAccelerator, systems are running linux 2.6.
almost running perfectly, from time to time the lightspeeds like to hang up suddenly, there is no reaction on all tcp ports, no output in any error logs, no core dumps, lightspeed just running doing nothing.
the init script for stopping is also not working, only -KILL on the lightscripts daemons do. after that i can restart normally. i noticed that "using vmstat", the buffered memory while using lightspeed never got freed, only after a complete reboot.
currently i have no straces of lightspeeds hanging, i will trigger the next time, maybe there is some useful information inside, but maybe u have another hint or suggestion how i can trace or even reproduce bug?
maybe there is another bug involed, that even without using the balancing system and without any load sometimes i got a tcp connection refused error.
thx for reading,
jason
i am currently using for a sport event 6 web nodes running lightspeed standard version 2.1.14 behind load balancers, serving around 800 requests/sec per node, writing 200/mb logfiles per node. php is also involed in using lsphp withj eAccelerator, systems are running linux 2.6.
almost running perfectly, from time to time the lightspeeds like to hang up suddenly, there is no reaction on all tcp ports, no output in any error logs, no core dumps, lightspeed just running doing nothing.
the init script for stopping is also not working, only -KILL on the lightscripts daemons do. after that i can restart normally. i noticed that "using vmstat", the buffered memory while using lightspeed never got freed, only after a complete reboot.
currently i have no straces of lightspeeds hanging, i will trigger the next time, maybe there is some useful information inside, but maybe u have another hint or suggestion how i can trace or even reproduce bug?
maybe there is another bug involed, that even without using the balancing system and without any load sometimes i got a tcp connection refused error.
thx for reading,
jason