Sat May 28 14:44:45 UTC 2005
My dccifd daemon has been hanging every few days. I think it started with 1.3.4 and it just happened again in 1.3.5. It's 100% cpu bound. I run it under daemontools as: libexec/dccifd -b -tCMN,40 -U userdirs \ -w whiteifd -S X-Habeas-SWE-3 -S List-ID I managed to get a gdb traceback before killing it, and it said; #0 <signal handler called> #1 0xa0088b25 in _syscall_sys_read () from /shlib/libc_s.4.3.0 #2 0x8049ba5 in tmp_read () #3 0x8049c43 in tmp_read_msg_in () #4 0x8049ea0 in hdrs_copy () #5 0x804a02c in body_copy () #6 0x804b597 in ascii_done () #7 0x804b9c8 in ascii_job () #8 0x804cdc9 in job_start () #9 0xa0036a2f in _thread_kern_start () from /shlib/libc_s.4.3.0 This suggests that whatever tmp_read is trying to read isn't there and it's not noticing that it's gone away. Anyone else see this? I'm rebuilding it with debug symbols so I can try and get more clues, but if anyone else has seen this and has some hints, I'd love to hear them. Regards, John Levine, email@example.com, Primary Perpetrator of "The Internet for Dummies", Information Superhighwayman wanna-be, http://iecc.com/johnl, Mayor "I dropped the toothpaste", said Tom, crestfallenly.
More information about the DCC