Bug #2041
closedmod_fastcgi chokes on v8cgi
Description
I tried to use v8cgi as fastcgi with lighttpd, both with spawnfcgi and without - both resulting in SIGUSR1 on worker start.
Compare with http://code.google.com/p/v8cgi/issues/detail?id=7&can=1 - Apache + mod_fcgi works without failure, so the issue could probably be in the mod_fastcgi of lighttpd.
Alas, the reason does not appear to be easily found: here's a diff between the straces of a working lua magnet worker and v8cgi:
< execve("/usr/local/bin/spawn-fcgi", ["spawn-fcgi", "-f", "/usr/bin/magnet", "-s", "/tmp/v8-socket", "-u", "andre"], [/* 21 vars */]) = 0
---
> execve("/usr/local/bin/spawn-fcgi", ["spawn-fcgi", "-f", "/usr/local/bin/v8cgi", "-s", "/tmp/v8-socket", "-u", "andre"], [/* 21 vars */]) = 0
64,65c64,65
< connect(3, {sa_family=AF_FILE, path="/tmp/v8-socket"}, 16) = -1 ENOENT (No such file or directory)
< unlink("/tmp/v8-socket") = -1 ENOENT (No such file or directory)
---
> connect(3, {sa_family=AF_FILE, path="/tmp/v8-socket"}, 16) = -1 ECONNREFUSED (Connection refused)
> unlink("/tmp/v8-socket") = 0
71,76c71,77
< clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7eb8ae8) = 1565
< select(0, NULL, NULL, NULL, {0, 100000}) = 0 (Timeout)
< waitpid(1565, 0xbfffd9d0, WNOHANG) = 0
< fstat64(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 2), ...}) = 0
< mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7fea000
< write(1, "spawn-fcgi.c.206: child spawned "..., 56) = 56
---
> clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7eb8ae8) = 1974
> select(0, NULL, NULL, NULL, {0, 100000}) = ? ERESTARTNOHAND (To be restarted)
> --- SIGCHLD (Child exited) @ 0 (0) ---
> select(0, NULL, NULL, NULL, {0, 91000}) = 0 (Timeout)
> waitpid(1974, [{WIFEXITED(s) && WEXITSTATUS(s) == 1}], WNOHANG) = 1974
> write(2, "spawn-fcgi.c.229: child exited w"..., 39spawn-fcgi.c.229: child exited with: 1
> ) = 39
79,80c80
< munmap(0xb7fea000, 4096) = 0
< exit_group(0) = ?
---
> exit_group(1) = ?
I would be really happy to get this working.
Greetings, LX
Updated by stbuehler almost 14 years ago
- Status changed from New to Invalid
Use the "-n" option for spawn-fcgi to debug fastcgi applications.
But this is a bug tracker, not a support forum.
And spawn-fcgi is not a part of lighttpd anymore, it is an extra package now.
Updated by LX almost 14 years ago
I will do so, yet the error happens without spawn-fcgi too, so I guess the problem lies rather with lighttpd than with spawn-fcgi. Once I know more, I will be back.
Updated by stbuehler almost 14 years ago
There is no FastCGI communication involved; v8cgi terminates before it has handled a single request, so the initialization failed.
The -n option may help to see errors on stderr (without -n the FastCGI application is detached from the terminal, so you don't see any errors), but you really should use spawn-fcgi 1.6.
Also available in: Atom