You are browsing a read-only backup copy of Wikitech. The live site can be found at wikitech.wikimedia.org
Problem determination and data collection
hhvm-dump-debug: captures a stack trace of HHVM and writes it to stdout and to
hhvm-dump-debug --fullto generate a core dump as well.
links -dump http://127.0.0.1/server-status -width 80: produce a full Apache status report, showing which requests are currently in flight.
furlis a cURL-like tool for FastCGI. You can use it to bypass Apache and request pages directly from HHVM. For example:
furl http://en.wikipedia.org/wiki/Main_Page. Requests always go to localhost.
curl localhost:9002/dump-apcwill dump all the current APC keys to
- stacktraces, if available, are in
- core dumps are in
Light process dying
HHVM runs requests in different threads but in a single process. fork() has to lock the entire memory space of the process so it can copy it, which blocks all requests that are currently in flight. If HHVM forks in advance, it can just have the child process sitting there, and then when a thread needs to shell out it can just call exec*.
If you see light processes deaths in the logs it could be because they always log when they exit (noisy but ignorable).
There are five pre-forked light processes so that worker threads, serving traffic, don't have to contend for one when they need an additional process forked.