I just noticed.. I made a script based on "skeleton" and I added; # Required-Start: $ALL when I executed insserv on the script it was linked as "S01script" and as such it's started first, yet the line above should tell insserv to isert the script to start LAST, not FIRST.. Ideas anyone? Anders.
Anders, On Sunday 13 February 2005 09:32, Anders Norrbring wrote:
I just noticed.. I made a script based on "skeleton" and I added;
# Required-Start: $ALL
when I executed insserv on the script it was linked as "S01script" and as such it's started first, yet the line above should tell insserv to isert the script to start LAST, not FIRST..
Ideas anyone?
Anders.
# strings /sbin/insserv |egrep -i '\$|all' malloc t/<$t $all $.#%_+-\*[]^:()~ -r, --remove Remove the listed scripts from all runlevels. ^(\$[a-z0-9_-]+)([[:blank:]]?[=:]+[[:blank:]]?|[[:blank:]]+)([[:print:][:blank:]]*) Perhaps the "$all" is not matched case-insensitively. Try "$all" instead of "$ALL". Randall Schulz
Anders,
On Sunday 13 February 2005 09:32, Anders Norrbring wrote:
I just noticed.. I made a script based on "skeleton" and I added;
# Required-Start: $ALL
when I executed insserv on the script it was linked as "S01script" and as such it's started first, yet the line above should tell insserv to isert the script to start LAST, not FIRST..
Ideas anyone?
Anders.
# strings /sbin/insserv |egrep -i '\$|all' malloc t/<$t $all $.#%_+-\*[]^:()~ -r, --remove Remove the listed scripts from all runlevels. ^(\$[a-z0-9_- ]+)([[:blank:]]?[=:]+[[:blank:]]?|[[:blank:]]+)([[:print:][:blank:]]*)
Perhaps the "$all" is not matched case-insensitively. Try "$all" instead of "$ALL".
Nope, same result if I use 'all' or 'ALL', no difference, it inserts it as S01... Anders.
On Sunday 13 February 2005 10:42 am, Anders Norrbring wrote:
Perhaps the "$all" is not matched case-insensitively. Try "$all" instead of "$ALL".
Nope, same result if I use 'all' or 'ALL', no difference, it inserts it as S01...
Maybe it needs to be defined in /etc/insserv.conf ? There's no '$all' there in my copy. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.8-24.11-default x86_64
On Sunday 13 February 2005 10:42 am, Anders Norrbring wrote:
Perhaps the "$all" is not matched case-insensitively. Try "$all"
instead
of "$ALL".
Nope, same result if I use 'all' or 'ALL', no difference, it inserts it as S01...
Maybe it needs to be defined in /etc/insserv.conf ? There's no '$all' there in my copy.
I doubt it, since it should be used to insert the script last of ALL scripts. Lets say I build a script that provides "Eggs"; ### BEGIN INIT INFO # Provides: Eggs Then I can add another script that depends on the Eggs; ### BEGIN INIT INFO # Provides: Chickens # Required-Start: Eggs When I do 'insserv chickens', it will be inserted to start after 'eggs'. insserv.conf doesn't need to have any knowledge about the eggs and chickens, if you look in it you'll see that $network is only short for the services 'network' 'pcmcia' and 'hotplug'. If you should define $ALL in there, you'd have to add every distribution script, PLUS all your own.. I think $ALL should just make insserv put your script at the highest possible S-value.. Anders.
On Sunday 13 February 2005 11:14 am, Anders Norrbring wrote:
If you should define $ALL in there, you'd have to add every distribution script, PLUS all your own.. I think $ALL should just make insserv put your script at the highest possible S-value..
Well, I did some googling and some reading of the LSB on boot facility names http://www.linuxbase.org/modules.php?name=specrev&url=http://www.linuxbase.org/spec//booksets/LSB-Core-generic/LSB-Core-generic.html I couldn't find any mention anywhere of $all as a valid facility. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.8-24.11-default x86_64
-----Original Message----- From: Scott Leighton [mailto:helphand@pacbell.net] Sent: Sunday, February 13, 2005 9:11 PM To: suse-linux-e@suse.com Subject: Re: [SLE] insserv broken?
On Sunday 13 February 2005 11:14 am, Anders Norrbring wrote:
If you should define $ALL in there, you'd have to add every distribution script, PLUS all your own.. I think $ALL should just make insserv put your script at the highest possible S-value..
Well, I did some googling and some reading of the LSB on boot facility names http://www.linuxbase.org/modules.php?name=specrev&url=http://www.linuxbase .org/spec//booksets/LSB-Core-generic/LSB-Core-generic.html
I couldn't find any mention anywhere of $all as a valid facility.
From the /etc/init.d/skeleton file:
# * When specifying hard dependencies or ordering requirements, you can # use names of services (contents of their Provides: section) # or pseudo names starting with a $. The following ones are available # according to LSB (1.1): # $local_fs all local file systems are mounted # (most services should need this!) # $remote_fs all remote file systems are mounted # (note that /usr may be remote, so # many services should Require this!) # $syslog system logging facility up # $network low level networking (eth card, ...) # $named hostname resolution available # $netdaemons all network daemons are running # The $netdaemons pseudo service has been removed in LSB 1.2. # For now, we still offer it for backward compatibility. # These are new (LSB 1.2): # $time the system time has been set correctly # $portmap SunRPC portmapping service available # UnitedLinux extensions: # $ALL indicates that a script should be inserted # at the end Apparently I overlooked that $ALL is a UnitedLinux extension, nevertheless I think it's kind of confusing to put it in the skeleton file for a non-UL distribution. My mistake.. Anders.
On Sunday 13 February 2005 1:29 pm, Anders Norrbring wrote:
From the /etc/init.d/skeleton file:
# UnitedLinux extensions: # $ALL indicates that a script should be inserted # at the end
Apparently I overlooked that $ALL is a UnitedLinux extension, nevertheless I think it's kind of confusing to put it in the skeleton file for a non-UL distribution.
So, does that mean it is or isn't available for SuSE. I admit, I'm confused about why the string $all would show up in insserv if it wasn't being used. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.8-24.11-default x86_64
Scott, On Sunday 13 February 2005 10:54, Scott Leighton wrote:
...
Maybe it needs to be defined in /etc/insserv.conf ? There's no '$all' there in my copy.
Look at the "strings" output I supplied. "$all" is compiled into insserv.
Scott
Randall Schulz
On Sunday 13 February 2005 1:29 pm, Randall R Schulz wrote:
Scott,
On Sunday 13 February 2005 10:54, Scott Leighton wrote:
...
Maybe it needs to be defined in /etc/insserv.conf ? There's no '$all' there in my copy.
Look at the "strings" output I supplied. "$all" is compiled into insserv.
But I thought the OP said he tried it and it didn't work. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.8-24.11-default x86_64
Scott, On Sunday 13 February 2005 13:54, Scott Leighton wrote:
On Sunday 13 February 2005 1:29 pm, Randall R Schulz wrote:
Scott,
On Sunday 13 February 2005 10:54, Scott Leighton wrote:
...
Maybe it needs to be defined in /etc/insserv.conf ? There's no '$all' there in my copy.
Look at the "strings" output I supplied. "$all" is compiled into insserv.
But I thought the OP said he tried it and it didn't work.
Yes. Yes to both. The string is there in the binary but Anders reports it not having the effect he expects. He also reported that according to comments in the config file that keyword is available only for United Linux (whatever that is). It's certainly not mentioned in the man page on my system.
Scott
Randall Schulz
Randall R Schulz
He also reported that according to comments in the config file that keyword is available only for United Linux (whatever that is).
On what planet have you been living? :))) UL was meant as a common Linux distributed and developed by SCO Group, Connectiva, SuSE and Turbo Linux. Now UL is long history. Whatever is documented as being part of UL is part of SuSE Linux so I'd rather call it a bug if it isn't working. Philipp
On Sunday 13 February 2005 18:32, Anders Norrbring wrote:
I just noticed.. I made a script based on "skeleton" and I added;
# Required-Start: $ALL
when I executed insserv on the script it was linked as "S01script" and as such it's started first, yet the line above should tell insserv to isert the script to start LAST, not FIRST..
Ideas anyone?
It works on my system. Example script evlog, default is Required-Start: network /etc/init.d # ls rc5.d |grep evlog K16evlog S06evlog insserv -r evlog, change to $all ($ALL gives the same result) and insserv evlog /etc/init.d # ls rc5.d |grep evlog K01evlog S21evlog S21 puts it dead last in my runlevel 5. What versions of things are you running?
On Sunday 13 February 2005 18:32, Anders Norrbring wrote:
I just noticed.. I made a script based on "skeleton" and I added;
# Required-Start: $ALL
when I executed insserv on the script it was linked as "S01script" and as such it's started first, yet the line above should tell insserv to isert the script to start LAST, not FIRST..
Ideas anyone?
It works on my system. Example script evlog, default is Required-Start: network
/etc/init.d # ls rc5.d |grep evlog K16evlog S06evlog
insserv -r evlog, change to $all ($ALL gives the same result) and insserv evlog /etc/init.d # ls rc5.d |grep evlog
K01evlog S21evlog
S21 puts it dead last in my runlevel 5.
What versions of things are you running?
I'm running SuSE Pro 9.1 stock+YOU updated. # insmod --version module-init-tools version 3.0-pre10 I did a new test... I made a script named "test", here are the results. ### BEGIN INIT INFO # Provides: Dummy # Required-Start: $all # ls rc3.d/ | grep test K21test S01test To me, this looks exactly REVERSED! Starts first, stops last.. *confused* I changed the script to: ### BEGIN INIT INFO # Provides: Dummy # Required-Start: $syslog Result: # ls rc3.d/ | grep test K14test S08test
S21 puts it dead last in my runlevel 5.
What versions of things are you running?
I'm running SuSE Pro 9.1 stock+YOU updated. # insmod --version module-init-tools version 3.0-pre10
Not that I know why I checked the version of insmod, must be lack of coffee.. Sorry. Anders.
participants (5)
-
Anders Johansson
-
Anders Norrbring
-
Philipp Thomas
-
Randall R Schulz
-
Scott Leighton