Anil wrote: > Apache goes unresponsive. I test using telnet to port 80 and do a GET, no > response comes back from the server. I then have to restart the service to > make it work. > > I used dtrace to capture some data and I see these calls: > ... > 3 83394 ___errno:entry > 3 80993 apr_atomic_dec32:entry > 3 83062 atomic_dec_32_nv:entry > 3 83394 ___errno:entry > 3 83394 ___errno:entry > 3 80853 apr_pollset_poll:entry > ... > > > Nothing else. I also ran into similar issues with Lighttpd14, though I did > not test it out with dtrace to see if it is behaving similar. Lighttpd14 is > on another zone, separate from the Apache install - but on the same 2009.06 > release. >
no clue here about Lighttpd > I saw another thread in this forum related to APR Event ports: > http://www.opensolaris.org/jive/thread.jspa?threadID=112432&tstart=0 > > Is that related? How can I get a work around to this? This makes the > apache/lighttpd very unreliable. Has anyone seen this? > It looks like the same problem given the available information. Please try the attached work-around and report back. The work-around consists of a library interposer which sits between APR and libc and makes port_getn() act like APR expects. A permanent solution is in development build 124 (not released yet) and future APR releases. -------------- next part -------------- A non-text attachment was scrubbed... Name: portfs_hang.tar.gz Type: application/x-gzip Size: 4470 bytes Desc: not available URL: <http://mail.opensolaris.org/pipermail/webstack-discuss/attachments/20090923/591b4ac3/attachment.bin>
