DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://nagoya.apache.org/bugzilla/show_bug.cgi?id=21649>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://nagoya.apache.org/bugzilla/show_bug.cgi?id=21649

apache restart fails with error "httpd: bad user name nobody"





------- Additional Comments From [EMAIL PROTECTED]  2003-07-28 13:32 -------
FYI...  There is another Apache function broken by this limitation of the 
Solaris library 
routines -- run-time lookup of home directories via mod_userdir.  There is an 
existing PR 
for this (I'll try to find later) claiming that there is a thread-safety issue. 
 The problem really 
is that when the server is more heavily loaded more file descriptors can be in 
use for 
those other active threads, and getpwnam() and friends can be more likely to 
fail.

I think what is needed is something at a lower-level to make the APR code more 
resiliant.  
The best compromise I can think of is reading the darn /etc/passwd ourselves 
with APR 
buffered I/O, which doesn't have the limitations of Solaris stdio w.r.t. file 
descriptor 
numbers.  We'd first have to verify that the system is using /etc/passwd 
(instead of NIS or 
LDAP or whatever), and use the normal API if not using /etc/passwd.  Butt ugly, 
but doing 
the fd slack stuff like Apache 1.3 is butt ugly too (trying to leave 
low-numbered fds 
available by duping descriptors from open() or accept() to a higher fd then 
closing the 
original one we got from open() or accept()); it has some thread issues to work 
out too.

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to