Thanks for your message! You describe precisely what happened when I removed
the non_volatile.db and the door file in /etc/svc/volatile yesterday.
On a test system I sent TERM to svc.startd and this launched a new svc.configd.
The svc system looks pretty scattered after that: legacy_run services
disappeared, other services in online, offline, maintaince or uninitialized
state. On the other hand /etc/inetd started up and worked correctly. After
rebooting the test system everything was find.
With this fresh experience I decided to do this on the production server.
Unfortunately things went a little bit different. TERM didn't terminate
svc.startd so (after some experiments on the test system) I sent a KILL (9) to
it. The states reported by svcs looks more or less like on the test system, but
inetd was not started. The reason is that /lib/svc/method/fs-usr feels that my
server's root-fs failed a sanity check:
[ Aug 24 10:08:57 Executing start method ("/lib/svc/method/fs-usr") ]
/dev/dsk/c0t0d0s1: Overlapping swap files are not allowed
/dev/dsk/c0t1d0s1: Overlapping swap files are not allowed
The / file system (/dev/md/rdsk/d30) is being checked.
/dev/md/dsk/d30: /dev/md/dsk/d30 IS CURRENTLY MOUNTED READ/WRITE.
/dev/md/dsk/d30: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.
WARNING - fatal error from fsck - error 33
Unable to repair the / filesystem. Run fsck manually
(fsck -F ufs /dev/md/rdsk/d30).
This worries me pretty much! Despite its lamentable state the machines executes
nfs, ssh and the stuff without any limitations. I announced a system
maintenance for next tuesday!
Do you have any recommendations for me? I must say, that my sympathies for the
Solaris Volume Manager is somewhat limited these days. A hard disk crash
shouldn't arise such a situation. Anyway there should be a way to salvage this
damn SQLite-file and get SMF in consistent state without a reboot.
Any hints or remarks are highly welcome! Thanks again, HJ
This message posted from opensolaris.org
