On Sun 18 Jan 2004, Gregor Zattler wrote: > > In last months a wwwoffle-htdig-full run completed in ca. 4 > hours. Now it hangs forever: > > # ps fax|grep tty6 > 561 tty6 S 0:02 -bash > 740 tty6 S 0:00 \_ /bin/sh search/htdig/scripts/wwwoffle-htdig-full > 741 tty6 R 1457:55 \_ htdig -i -c search/htdig/conf/htdig-full.conf
In such a case, check what files it has open: # lsof -p 741 You can also see what it's doing: # strace -p 741 lots of output from this one, probably... if it shows nothing, tnen it's hanging in user space, i.e. not doing any system calls. In that case it's a definite bug in htdig, # ltrace -p 741 might be useful. If that show nothing as well, then it's spinning somewhere in a non-stop loop. > But the files in /var/cache/wwwoffle/search/htdig/db were last updated > more than 24 hours ago: > > # ls -Al /var/cache/wwwoffle/search/htdig/db > total 378612 > -rw-r--r-- 1 root root 179151872 2004-01-17 16:11 db.docdb > -rw-r--r-- 1 root root 208151269 2004-01-17 16:11 db.wordlist It could be that the format of those files changed, making htdig go into the loop when it tries to update them. Start fresh... > P.S.: > It's not a problem of ressources: > > top - 18:09:13 up 2 days, 6:10, 9 users, load average: 2.00, 2.00, 2.22 > Tasks: 83 total, 8 running, 75 sleeping, 0 stopped, 0 zombie > Cpu(s): 99.9% user, 0.1% system, 0.0% nice, 0.0% idle > Mem: 515940k total, 505904k used, 10036k free, 80872k buffers > Swap: 875500k total, 181708k used, 693792k free, 84192k cached > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 741 root 20 0 23116 732 676 R 49.9 0.1 1457:41 htdig > 1791 root 20 0 7560 7288 608 R 49.9 1.4 32:09.04 dpkg Hmm, why is dpkg also eating lots of time?! Paul Slootman
