Opened 23 years ago
Closed 21 years ago
#325 closed defect (fixed)
lfs-bootscripts-1.13
Reported by: | Owned by: | ||
---|---|---|---|
Priority: | normal | Milestone: | |
Component: | Book | Version: | CVS |
Severity: | normal | Keywords: | |
Cc: | list@… |
Description
Use variables in the daemon start scripts, like:
source /etc/sysconfig/rc source $rc_functions
NAME="" #name of the service/application PREFIX="" #path to the application APP="" #the application OPTIONS="" #parameters
case "$1" in
start)
echo "Starting $NAME..." loadproc $PREFIX/$APP $OPTIONS ;;
and so on
Attachments (9)
Change History (33)
comment:1 by , 23 years ago
dependson: | → 30 |
---|
comment:2 by , 23 years ago
by , 23 years ago
Fixed the priority digit stripping to obtain $suffix - will work with mixed amount of numbers
comment:4 by , 23 years ago
Write more docs regarding usage of the scripts, especially the network script's ability to use /etc/sysconfig/network-devicese/ifup-devicename script to do the work rather than the default ifconfig command run by the ifup script (same for ifdown-devicename)
comment:5 by , 23 years ago
Instead of just shutting down on error, see what can be done to ask the user what to do when an error has been encountered.
comment:7 by , 23 years ago
On lfs-dev, Erika Pacholleck pointed out that the above change to use find:
Needs (at least here, dunno about standard lfs) mv /usr/bin/find /bin
so if it's decided to do this, we'll need to add the relevant mv to chapter6 findutils.
comment:8 by , 23 years ago
From: list@…
I found following bug in /etc/rc.d/init.d/functions. If you use a serial console "stty size" (line 16) always reports "0 0". When you start something like "/etc/rc.d/init.d/sysklogd start" the "[ OK ]" message beginns at column 0 and overwrites the "starting ..." text. Following patch resolves this problem. ============== snip ==================== --- functions.orig 2002-05-29 17:07:31.000000000 +0200 +++ functions 2002-05-29 17:09:10.000000000 +0200 @@ -15,6 +15,10 @@
COLUMNS=$(stty size) COLUMNS=${COLUMNS##* }
+ +# on serial console stty reports 0 +# change this to 80 here +if [ "$COLUMNS" == "0" ] +then + COLUMNS=80 +fi
COL=$[ $COLUMNS - 10 ] WCOL=$[ $COLUMNS - 30 ] SET_COL="echo -en
033[${COL}G"
============== /snip ====================
comment:10 by , 22 years ago
Priority: | normal → highest |
---|
comment:11 by , 22 years ago
Addition to comment #7 - don't just run 'find' (if we are going to do it, not sure yet) on /var/run also run it on /var/lock
comment:12 by , 22 years ago
Priority: | highest → lowest |
---|---|
Summary: | lfs-bootscripts-1.10 → lfs-bootscripts-1.11 |
I am releasing 1.10 now but nothing much has changed other than a small change so checkfs can work with non-ext2 file systems better. All the changes here will go into 1.11 in the future (too much work to test all the new code).
comment:13 by , 22 years ago
Another thing you might want to consider adding would be to set the domainname correctly:
Maybe add a line to /etc/sysconfig/network DOMAINNAME=linuxfromscratch.org
And then apply this patch to localnet
--- /home/mark/LFS/LFS/bootscripts/rc.d/init.d/localnet 2002-03-16 18:26:44.000000000 +0000 +++ /etc/rc.d/init.d/localnet 2002-09-17 20:12:19.000000000 +0100 @@ -15,6 +15,9 @@
echo "Setting hostname to $HOSTNAME..." loadproc hostname $HOSTNAME
+ + echo "Setting domainname to $DOMAINNAME..." + loadproc domainname $DOMAINNAME
;;
stop)
I don't know why, it just feels cleaner that when I run domainname, it gives me the proper answer ;-) (Oh, and I think it should go in the base bootscripts because I consider it part of a basic network initialisation procedure).
by , 22 years ago
Attachment: | LFS-norundel-2002.12.24.12.patch added |
---|
Don't remove /var/run/* instead clean it out, leaving directories intact
by , 22 years ago
Attachment: | lfs-bootscripts-1.10-syntax.patch added |
---|
Patch to make the scripts compatible with ash from busybox
comment:14 by , 22 years ago
Priority: | lowest → highest |
---|
by , 22 years ago
Attachment: | lfs-bootscripts-1.11-fixes.diff added |
---|
comment:15 by , 22 years ago
Summary: | lfs-bootscripts-1.11 → lfs-bootscripts-1.12 |
---|
comment:16 by , 22 years ago
See http://archive.linuxfromscratch.org/mail-archives/lfs-dev/2003/05/0980.html
The idea is to make the network scripts more generic as to allow different services (ie: DHCP instead of static IP only as assumed now) to be plugged in without the need to rewrite a bunch of files.
This should be tested before implementing.
comment:17 by , 22 years ago
attachments.isobsolete: | 0 → 1 |
---|
by , 22 years ago
Attachment: | network.tar.bz2 added |
---|
Fixed version of conathan's updated network scripts.
comment:18 by , 21 years ago
There is a problem with fsck return codes that needs fixing. See the thread here for details:-
http://linuxfromscratch.org/pipermail/lfs-dev/2003-August/036184.html
I'll attach a suggested patch.
comment:19 by , 21 years ago
Priority: | highest → high |
---|
comment:21 by , 21 years ago
dependson: | 30 |
---|
comment:22 by , 21 years ago
Bootscripts currently use "touch" unnecessarily.
It can be replaced by "> /var/run/utmp" like is currently done in the mountfs script for /etc/mtab.
See Bug 679 for some background.
(we can stop moving "touch" into /bin once the bootscripts are modified).
comment:23 by , 21 years ago
Priority: | high → normal |
---|
comment:24 by , 21 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
Closing this bug for LFS-bootscripts for now - bootscript problems are now tracked as part of the LFS-bootscripts bugzilla project.
When collecting K* links, there's a bug that causes you not being able to use more than two digits for the links, else you'll get errors that there is not corresponding S* link in the previous level.
See discussion started:
From: ask@… (DaClink) Subject: An Annoying message when switching runlevels Date: Thu, 11 Apr 2002 10:57:12 +0000 (UTC) To: blfs-support@…