On Thu, Jan 17, 2013 at 11:07 PM, Ben Noordhuis <i...@bnoordhuis.nl> wrote: > On Thu, Jan 17, 2013 at 8:37 PM, Chris Scribner <scr...@gmail.com> wrote: >> Thanks for the information so far. I'll try it out and see what sticks. >> >> A little more information... >> >> I was running a 1 master + 3 worker cluster (using node's clustering) when >> the profile was taken. Do you know how the profile combines results from the >> various processes? For instance, would it make sense that the >> libsystem_kernel was for the master process that wasn't doing much? > > Yes, quite possibly, or maybe the other way around - from one of the workers. > > The --prof flags gets passed on to the workers but the problem is that > it either creates a garbage v8.log (because it contains entries from > multiple processes) or that only one process actually gets to write to > it, the last one that opens it. > > What you can do as a workaround is to pass `--logfile v8-%t.log` on > the command line and make sure that each worker is started with at > least a one millisecond delay. %t gets replaced with the timestamp in > milliseconds.
Sorry, I have a typo in there: it's --logfile=v8-%t.log (notice the equals sign). I think I'll land a patch in node that automatically appends the PID to the file name when in cluster mode. -- Job Board: http://jobs.nodejs.org/ Posting guidelines: https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines You received this message because you are subscribed to the Google Groups "nodejs" group. To post to this group, send email to nodejs@googlegroups.com To unsubscribe from this group, send email to nodejs+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/nodejs?hl=en?hl=en