This page validates as XHTML 1.0, and will look much better in a browser that supports web standards, but it is accessible to any browser or Internet device. It was created using techniques detailed at glish.com/css/.
Tuesday April 12, 2005 Nice walk in Farnham Park this evening, whole family seemed to enjoy it. We disturbed some of the deer luckily the samoyeds didn't spot them.
Customer installed QFS , rebooted and then QFS wouldn't start, complained it could not get a major number for qfsioc, remove and reload the package and it would work until the next reboot. Much debugging had gone on to no avail. The answer came from the explorer run that had been requested. The systems OBP boot device pointed towards the /devices path for c0t0d0s0, but the system had a triple mirrored root device using disksuite. The explorer showed that the mirror that occupied c0t0s0s0 was in maintenance state. So the booting machine accesses the physical c0t0d0s0 device and reads the bootblock , /etc/system and several other kernel files and then loads the required drivers and mounts the pseudo sds root device . Suddenly any updates will go to the surviving two mirrors and not to the c0t0d0s0 copy. So the QFS filesystem drivers had been applied to the pseudo sds root filesystem but not to the frozen c0t0d0s0 filesystem. Hopefully all should be well when the needs maintenance state is resolved.
Making progress on the syseventd escalation, two days reading the code has allowed me to propose a simple experiment to see if the problem is a simple race between syseventd starting devfsadmd and devfsadmd getting to the point where it can subscribe to get events. Lets see what is in the output of the ..
truss -d -D -fl -o /var/tmp/out /usr/lib/sysevent/syseventd
15 miles in the smart - legs ached.
0 miles on the bike.