Yep, I just got hit by the runaway syslogd process on my Quad G5. I use Temperature Monitor so I could see the records of my CPU temp going up for long periods of time. It had such high CPU utilization, it made the fans run higher, syslogd seemed to run for hours and I sure don't want to put a strain on the old cooling system. I can't imagine what this would do to battery life on a laptop Mac.
But I saw another fix. I guess we should do this through official methods, use lanchctl to stop it and then delete the database (which can grow quite large, it will be recreated on startup). Use these terminal commands:

sudo launchctl stop com.apple.syslogd
sudo rm /var/log/asl.db

You can restart syslogd:

sudo launchctl start com.apple.syslogd

Once the daemon is stopped and the DB is purged, it seems to run stably when you restart it. According to a comment here, where I found this solution, this is logged as a bug with Apple and hopefully there will be a fix in an update.

Now if only I could figure out why I get runaway Finder processes. It puts just as big a load on the CPU as syslogd did.

[ Reply to This | # ]

I found I had to nuke my system.log file as well, which was over 5GB...since then, syslogd is behaving (well, for at least 15 minutes...)

---
Microsoft MVP for Entourage
AppleScripts for OE and Entourage

[ Reply to This | # ]

I tried that solution as well. It seems to working. Personally my problem seemed to be caused by the program managing the HSDPA modem. It is logging as crazy indeed. So I did two things:

- In /etc/syslog.conf "*.notice" now logs to /dev/null
- I stopped syslogd with launchctl, removed /var/log/asl.db, relaunched syslogd

Now the CPU consumption of syslogd is near to zero, but neither the console nor asl.db are logging anything. I hope it is the regular behavior.


[ Reply to This | # ]

I can also confirm that zapping the file asl.db fixed the problem. My file was about 2 months old and had grown to around 250Mb. The system log file was full of three line messages varying from seconds to a minute apart telling me that syslogd had a memory error.

com.apple.syslogd[1248]: syslogd(1248,0xa0750fa0) malloc: *** mmap(size=3095314432) failed (error code=12)
com.apple.syslogd[1248]: *** error: can't allocate region
com.apple.syslogd[1248]: *** set a breakpoint in malloc_error_break to debug

One CPU has spent the last three months maxed out at 100%. I fixed this three days ago after 10.5.2 didn't fix it and I found this thread and syslogd has not complained since.

I also notice that the asl.db clears itself frequently now. The current file is less than 24 hours old so some daily task is doing its required cleanup job.

Happy again.....

[ Reply to This | # ]