Bug 32687
| Summary: | scripts in /usr/local/etc/rc.d not executed in 4.4-STABLE | ||
|---|---|---|---|
| Product: | Base System | Reporter: | Nicholas Paufler <echofox> |
| Component: | misc | Assignee: | Peter Pentchev <roam> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | 4.4-STABLE | ||
| Hardware: | Any | ||
| OS: | Any | ||
On Mon, Dec 10, 2001 at 08:58:06PM -0800, Nicholas Paufler wrote:
>
> >Number: 32687
> >Category: misc
> >Synopsis: scripts in /usr/local/etc/rc.d not executed in 4.4-STABLE
> >Originator: Nicholas Paufler
> >Release: FreeBSD 4.4-STABLE #0: Mon Dec 10 18:49:42 MST 2001
> >Organization:
> >Environment:
> FreeBSD www.discordia.ca 4.4-STABLE FreeBSD 4.4-STABLE #0: Mon Dec 10 18:49:42 MST 2001 root@www.discordia.ca:/usr/obj/usr2/src/sys/DISCORDIA i386
> >Description:
> After doing an upgrade from 4.3-STABLE (built May 08 2001) to 4.4-STABLE (built Dec 10 2001) I noticed that local packages being started via scripts in /usr/local/etc/rc.d were not being executed upon reboot.
> I'd done a mergemaster after completing the make installworld in which I installed all new files except for rc.conf and rc.firewall.
OK, dumb question: does your /etc/defaults/rc.conf have a line like:
script_name_sep=" " # Change if your startup scripts' names contain spaces
G'luck,
Peter
--
What would this sentence be like if it weren't self-referential?
Forwarding a message from the submitter to put it on the audit trail.. G'luck, Peter -- If the meanings of 'true' and 'false' were switched, then this sentence wouldn't be false. ----- Forwarded message from Nicholas Echo|Fox Paufler <echofox@discordia.ca> ----- Date: Tue, 11 Dec 2001 09:04:36 -0700 (MST) From: "Nicholas \"Echo|Fox\" Paufler" <echofox@discordia.ca> X-Sender: echofox@kindercampus.ab.ca To: Peter Pentchev <roam@ringlet.net> Subject: Re: misc/32687: scripts in /usr/local/etc/rc.d not executed in 4.4-STABLE In-Reply-To: <20011211175055.A578@straylight.oblivion.bg> > > OK, dumb question: does your /etc/defaults/rc.conf have a line like: > script_name_sep=" " # Change if your startup scripts' names contain spaces > > G'luck, > Peter > Nope, that it does not. ... Ahhh.. just looking at the line now. I get what the problem is. since script_name_sep is undefined, slist is going to be equal to something like "apache.shmysql.shdhcpd.sh" and thus isn't going to exceute them because obviously such a file does not exist. So I take it my problem is an old and crufty version of /etc/defaults/rc.conf, then? Hmm. I would have thought mergemaster would have installed a current one, especially in a case where stuff is dependent on new additions to it. I trust that copying over /usr/src/etc/defaults/rc.conf should take care of my problems? Thanks for your help. Nicholas Paufler ----- End forwarded message ----- State Changed From-To: open->closed It turns out that the mergemaster run had for some reason not updated /etc/defaults/rc.conf. On Tue, 11 Dec 2001 08:20:01 PST, Peter Pentchev wrote:
> Ahhh.. just looking at the line now. I get what the problem is. since
> script_name_sep is undefined, slist is going to be equal to something like
> "apache.shmysql.shdhcpd.sh" and thus isn't going to exceute them because
> obviously such a file does not exist.
Peter, do you think it's worth building a default value for script_name_sep
into rc itself so this doesn't bite folks in the 4.5-RELEASE upgrade?
I accept that you'd have to try hard to be bitten, but this little bit
of ugliness may save us a few PRs?
Ciao,
Sheldon.
State Changed From-To: closed->analyzed Actually, Sheldon suggested a helpful workaround. On Tue, Dec 11, 2001 at 08:30:01AM -0800, Sheldon Hearn wrote:
> The following reply was made to PR misc/32687; it has been noted by GNATS.
>
> On Tue, 11 Dec 2001 08:20:01 PST, Peter Pentchev wrote:
>
> > Ahhh.. just looking at the line now. I get what the problem is. since
> > script_name_sep is undefined, slist is going to be equal to something like
> > "apache.shmysql.shdhcpd.sh" and thus isn't going to exceute them because
> > obviously such a file does not exist.
>
> Peter, do you think it's worth building a default value for script_name_sep
> into rc itself so this doesn't bite folks in the 4.5-RELEASE upgrade?
>
> I accept that you'd have to try hard to be bitten, but this little bit
> of ugliness may save us a few PRs?
This makes sense. How about the attached patch?
The warning message clutters the output, though; it almost certainly
overflows to the next line on the screen, not to mention overflowing
the 80-character source line.. I am thinking of shortening it somehow,
maybe you or somebody else could suggest another wording?
G'luck,
Peter
--
I am the meaning of this sentence.
Index: src/etc/rc
===================================================================
RCS file: /home/ncvs/src/etc/rc,v
retrieving revision 1.287
diff -u -r1.287 rc
--- src/etc/rc 11 Dec 2001 08:21:45 -0000 1.287
+++ src/etc/rc 11 Dec 2001 16:43:24 -0000
@@ -813,6 +813,10 @@
*)
echo -n 'Local package initialization:'
slist=""
+ if [ -z "${script_name_sep}" ]; then
+ echo -n ' Empty script_name_sep, setting a default value of " "..'
+ script_name_sep=" "
+ fi
for dir in ${local_startup}; do
if [ -d "${dir}" ]; then
for script in ${dir}/*.sh; do
Index: src/etc/rc.shutdown
===================================================================
RCS file: /home/ncvs/src/etc/rc.shutdown,v
retrieving revision 1.20
diff -u -r1.20 rc.shutdown
--- src/etc/rc.shutdown 24 Nov 2001 16:12:03 -0000 1.20
+++ src/etc/rc.shutdown 11 Dec 2001 16:42:36 -0000
@@ -103,6 +103,10 @@
;;
*)
slist=""
+ if [ -z "${script_name_sep}" ]; then
+ echo -n ' Empty script_name_sep, setting a default value of " "..'
+ script_name_sep=" "
+ fi
for dir in ${local_startup}; do
if [ -d "${dir}" ]; then
for script in ${dir}/*.sh; do
On Tue, 11 Dec 2001 18:51:49 +0200, Peter Pentchev wrote:
> This makes sense. How about the attached patch?
> The warning message clutters the output, though; it almost certainly
> overflows to the next line on the screen, not to mention overflowing
> the 80-character source line.. I am thinking of shortening it somehow,
> maybe you or somebody else could suggest another wording?
I'd suggest removing the warning altogether. Anyone who's managed to
mangle things in a 4.5-RELEASE upgrade to the point where they see this
message is unlikely to know how to make it go away.
I like the rest of the patch. :-)
Ciao,
Sheldon.
On Tue, Dec 11, 2001 at 07:00:07PM +0200, Sheldon Hearn wrote:
>
>
> On Tue, 11 Dec 2001 18:51:49 +0200, Peter Pentchev wrote:
>
> > This makes sense. How about the attached patch?
> > The warning message clutters the output, though; it almost certainly
> > overflows to the next line on the screen, not to mention overflowing
> > the 80-character source line.. I am thinking of shortening it somehow,
> > maybe you or somebody else could suggest another wording?
>
> I'd suggest removing the warning altogether. Anyone who's managed to
> mangle things in a 4.5-RELEASE upgrade to the point where they see this
> message is unlikely to know how to make it go away.
>
> I like the rest of the patch. :-)
Okay; the warning was actually an afterthought :)
If nobody objects, I will commit the attached patch tomorrow morning
(my morning, about twelve hours from now :)
G'luck,
Peter
--
If the meanings of 'true' and 'false' were switched, then this sentence wouldn't be false.
Index: src/etc/rc
===================================================================
RCS file: /home/ncvs/src/etc/rc,v
retrieving revision 1.287
diff -u -r1.287 rc
--- src/etc/rc 11 Dec 2001 08:21:45 -0000 1.287
+++ src/etc/rc 11 Dec 2001 17:04:51 -0000
@@ -813,6 +813,9 @@
*)
echo -n 'Local package initialization:'
slist=""
+ if [ -z "${script_name_sep}" ]; then
+ script_name_sep=" "
+ fi
for dir in ${local_startup}; do
if [ -d "${dir}" ]; then
for script in ${dir}/*.sh; do
Index: src/etc/rc.shutdown
===================================================================
RCS file: /home/ncvs/src/etc/rc.shutdown,v
retrieving revision 1.20
diff -u -r1.20 rc.shutdown
--- src/etc/rc.shutdown 24 Nov 2001 16:12:03 -0000 1.20
+++ src/etc/rc.shutdown 11 Dec 2001 17:04:52 -0000
@@ -103,6 +103,9 @@
;;
*)
slist=""
+ if [ -z "${script_name_sep}" ]; then
+ script_name_sep=" "
+ fi
for dir in ${local_startup}; do
if [ -d "${dir}" ]; then
for script in ${dir}/*.sh; do
Responsible Changed From-To: freebsd-bugs->roam Peter's gonna deal with this one. State Changed From-To: analyzed->suspended I committed the fix to -current, will MFC it in a week. State Changed From-To: suspended->closed The initialization of script_name_sep with a sane value was MFC'd into -stable yesterday. Thank you for the problem report! |
After doing an upgrade from 4.3-STABLE (built May 08 2001) to 4.4-STABLE (built Dec 10 2001) I noticed that local packages being started via scripts in /usr/local/etc/rc.d were not being executed upon reboot. I'd done a mergemaster after completing the make installworld in which I installed all new files except for rc.conf and rc.firewall. I eventually traced the problem to /etc/rc For whatever reason the local package initialization section of that file was failing to function correctly. I edited that file and pasted in the section from a different 4.3-STABLE machine, rebooted and the problem was solved. All local packages (apache, dhcpd, mysql) started properly. Fix: In the /etc/rc for 4.4-STABLE as of December 10th: Replace the current section: case ${local_startup} in [Nn][Oo] | '') ;; *) echo -n 'Local package initialization:' slist="" for dir in ${local_startup}; do if [ -d "${dir}" ]; then for script in ${dir}/*.sh; do slist="${slist}${script_name_sep}${script}" done fi done script_save_sep="$IFS" IFS="${script_name_sep}" for script in ${slist}; do if [ -x "${script}" ]; then (set -T trap 'exit 1' 2 ${script} start) fi done IFS="${script_save_sep}" echo '.' ;; esac With an earlier version of that section, such as the following taken from a 4.3-STABLE system: case ${local_startup} in [Nn][Oo] | '') ;; *) echo -n 'Local package initialization:' for dir in ${local_startup}; do if [ -d "${dir}" ]; then for script in ${dir}/*.sh; do if [ -x "${script}" ]; then (set -T trap 'exit 1' 2 ${script} start) fi done fi done echo '.' ;; esac I'm guessing that there is a typo or some other problem in the local package init section of the current /etc/rc which is causing the problem. How-To-Repeat: Local packages will not be initialized until that section of /etc/rc is replaced so every time the computer reboots the problem will repeat.