xref: /illumos-gate/usr/src/cmd/svc/milestone/README.share (revision 654b400c)

#
# CDDL HEADER START
#
# The contents of this file are subject to the terms of the
# Common Development and Distribution License (the "License").
# You may not use this file except in compliance with the License.
#
# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
# or http://www.opensolaris.org/os/licensing.
# See the License for the specific language governing permissions
# and limitations under the License.
#
# When distributing Covered Code, include this CDDL HEADER in each
# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
# If applicable, add the following below this CDDL HEADER, with the
# fields enclosed by brackets "[]" replaced with your own identifying
# information: Portions Copyright [yyyy] [name of copyright owner]
#
# CDDL HEADER END
#

Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
Use is subject to license terms.

/lib/svc/share/README

smf(5):  Notes on maintenance mode and recovery

Failures that bring the system to maintenance mode may include hardware
or critical software failures.  The procedures below are given so that
some software repairs can be made; the recommended exit approach once a
repair has been made is to reboot the system.  The system can be brought
to maintenance mode deliberately via the '-s' option to boot(1M), or via
the 's' option to init(1M).

In failure scenarios, smf(5) may or may not be running, depending on
which component has failed.  If smf(5) is running, and the /usr
filesystem is reachable, then the usual svcadm(1M) invocations to clear
maintenance state and restart services instances can be used.
Otherwise, the following instructions describe the direct execution of
service methods, so that capabilities that svc.startd(1M) would normally
start automatically can be started manually.  In the case that the
document recommends an invocation like

# /lib/svc/method/example-method start

you may also consider running these scripts with the shell displaying
the commands from the service method as they are executed.  For sh(1)
based scripts, this would mean running the method as

# /sbin/sh -x /lib/svc/method/example-method start

Some methods may be written to instead use ksh(1), with invocation

# /usr/bin/ksh -x /lib/svc/method/example-method start

The first line of the service method script will generally specify its
required interpreter using the standard #! notation.  Method scripts may
potentially require interpreters other than sh(1) or ksh(1).

1.  Boot archive failure

The boot archive may become out of sync with the root filesystem in a
reboot following an abnormal system shutdown. The recommended action is
to reboot immediately to rebuild the archive and correct the inconsistency.
To accomplish this, on a GRUB-based platform, choose "Solaris failsafe"
when the boot menu is displayed.  Type 'i' to get an interactive recovery
shell and follow instructions to update the boot archive.  On an OBP-
based platform, type 'boot -F failsafe' and follow the instructions.

If the list of stale files are not yet loaded by the kernel
or are compatible, you may continue booting by clearing the
boot-archive service state

# svcadm clear system/boot-archive

2.  Failure to mount filesystems.

In cases where the system was unable to bring a combination of the
system/filesystem/{root,usr,minimal} services online, it may be possible
to directly execute the corresponding service methods

# /lib/svc/method/fs-root
# /lib/svc/method/fs-usr
# /lib/svc/method/fs-minimal

to mount the various filesystems.  In the case that these methods fail,
a direct invocation of mount(1M), and potentially fsck(1M), should be
attempted for file systems required for recovery purposes.

/lib/svc/method/fs-usr attempts to remount the root file system
read-write, such that persistent changes can be made to the system's
configuration.  If this method is failing, one can directly remount
using the mount(1M) command via

# /sbin/mount -o rw,remount /

/etc/svc/volatile is a temporary filesystem generally reserved for Sun
private use.  It may prove a useful location to create mount points if
the root file system cannot be remounted read-write.

3.  Failure to run svc.configd(1M).

svc.configd(1M) will give detailed instructions for recovery if the
corruption is detected in the repository.  If svc.configd(1M) cannot be
run because of missing or corrupt library components, then the affected
components will need to be replaced.  Components could be copied from a
CD-ROM or DVD-ROM, or from another system.

4.  Failure to run svc.startd(1M).

If the inittab(4) line to invoke svc.startd(1M) is missing or incorrect,
it will need to be restored.  A valid entry is

smf::sysinit:/lib/svc/bin/svc.startd    >/dev/msglog 2<>/dev/msglog </dev/console

If svc.startd(1M) cannot be run because of missing or corrupt library
components, then the affected components will need to be replaced, as
for svc.configd(1M) above.

5.  Activating basic networking configuration.

If svc.startd(1M) did not execute successfully, it may also be necessary
to activate network interfaces manually, such that other hosts can be
contacted.  The service methods can be invoked directly as

# /lib/svc/method/net-loopback
# /lib/svc/method/net-physical

If these methods fail, a direct invocation of ifconfig(1M) can be
attempted.

In some scenarios, one may be able to use routeadm(1M) to activate more
dynamic route management functionality; restoring the default dynamic
routing behaviour can be done using the '-u' option.  (Invoking routeadm
with no arguments will display which commands must be accessible for the
current routing configuration to be invoked.)  Otherwise, once
interfaces are up, a default route can be manually added using the
route(1M) command.  On typical IPv4 networks, this invocation would be

# /sbin/route add net default _gateway_IP_

--

(An extended version of this document is available at
http://illumos.org/msg/SMF-8000-QD.  That version includes additional
document references.)