Folks,
This may be just something wrong in our environment, but it seems like we are seeing more and more Solaris 10 systems that need a fsck on reboot even if the shutdown was (apparently) clean. Usually we just need to login to maintenance mode and run "fsck -y /" and all is well, but occasionally we get stung by bug 6475797 and the admin has to clear a few inodes before fsck can work properly.
We've seen these on reboots just to restart applications (yea, I know...very "windows"), though most often they are seen on reboots following patching (from single user mode). In all cases the shutdown process appears clean and correctly executed.
It is always the root partition that's the problem (we use a single big root partition here, no separate /var, /usr, etc., though /home is on it's own). We've tried turning logging off (no difference) and verified that our shutdown script is just running the following command:
/usr/sbin/shutdown -y -i 6 -g 180
I don't expect folks to "solve" this for us, but I wanted to check out in the wider world to see if we're not alone...
Thanks!