You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jeff Trawick <tr...@attglobal.net> on 2002/05/15 20:17:55 UTC
is httpd a valid way to start Apache?
(of course this is a loaded question)
some problems with invoking httpd directly with Apache 2.0:
a. from an Apache binary build:
busted, since it won't find libapr et al (if you use a
Bourne-compatible shell you can source the envvars
file and it will work)
b. from a build done locally from source:
httpd loads fine but any special environment variable settings from
envvars are lost; the examples in CVS currently are that on AIX and
OS/390 Apache won't be able to get many threads without the envvar
override (as above, if you use a Bourne-compatible shell you can
source the envvars file and it will work)
The apxs and apachectl shell scripts deal with this by sourcing the
envvars file (that's why it is there, actually). If httpd were a
shell script that sourced the envvars file and ran something else
(e.g., httpd.bin) then apxs and apachectl wouldn't have to do any
dirty work and direct invocations of httpd would work just as well as
apachectl.
#!/bin/sh
. sbindir/envvars
sbindir/httpd.bin $*
It is ugly for httpd to be a shell script, but it is also ugly to get
different results based on how Apache was built and what OS you're on
and so forth.
If httpd is to be left as-is on Unix (i.e., maybe it works maybe it
doesn't), then perhaps apachectl needs to be pushed as the way to
access all httpd functionality.
(I loathe this stuff, but is going to be more important as 2.0.x
settles in and we have more users of binary builds.)
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Manoj Kasichainula <ma...@io.com>.
On Thu, May 16, 2002 at 05:00:13PM +0200, Jeroen Massar wrote:
> Due to inheritance (export) of environment variables I usually start
> Apache after doing a:
> # for i in `export | cut -f3 -d' '|cut -f1 -d'='`; do export -n $i; done
> Which cleans them all up nicely.
You can also do this by running httpd under "env -", which I believe
is quite portable. man env(1) for details.
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
"Ryan Bloom" <rb...@covalent.net> writes:
> My own opinion is that we leave things exactly as they are today. If
> you are running the binary by hand, you are taking some responsibility
> for knowing what you are doing. That means having the environment
> variables setup correctly before you start.
>
> If you don't want that responsibility, use apachectl to run the server.
> Trying to solve this problem any other way just seems like we are asking
> for trouble.
One aspect of these changes is that they resolve existing problems
with apachectl:
. apachectl doesn't give you full access to everything that you need
(all command-line options)
. apachectl doesn't take into account the possible movement of the pid
file
I don't think we can truthfully tell Joe User "use apachectl and
everything will work." You can't even get the server version out of
apachectl.
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
RE: is httpd a valid way to start Apache?
Posted by Scott Hess <sc...@avantgo.com>.
On Thu, 16 May 2002, Joshua Slive wrote:
> On Thu, 16 May 2002, Ryan Bloom wrote:
> > My own opinion is that we leave things exactly as they are today. If
> > you are running the binary by hand, you are taking some responsibility
> > for knowing what you are doing. That means having the environment
> > variables setup correctly before you start.
> >
> > If you don't want that responsibility, use apachectl to run the
> > server. Trying to solve this problem any other way just seems like we
> > are asking for trouble.
>
> I think that is exactly what this proposal is saying. But at the same
> time it is cleaning up apachectl and adding some useful functionality to
> httpd. As I've said, the current apachectl is over-complicated and the
> split between apachectl and httpd is confusing to some people. This
> change would clear that up.
Would it make sense to move the httpd binary to .../libexec/httpd? That
makes it clear that this is an internal binary which you shouldn't run
directly, unless you're really smart. Then apachectl stays in .../sbin/.
[Idea courtesy of mysql's mysqld.]
Later,
scott
RE: is httpd a valid way to start Apache?
Posted by Rasmus Lerdorf <ra...@apache.org>.
> > My own opinion is that we leave things exactly as they are today. If
> > you are running the binary by hand, you are taking some responsibility
> > for knowing what you are doing. That means having the environment
> > variables setup correctly before you start.
> >
> > If you don't want that responsibility, use apachectl to run the server.
> > Trying to solve this problem any other way just seems like we are asking
> > for trouble.
>
> I think that is exactly what this proposal is saying. But at the same
> time it is cleaning up apachectl and adding some useful functionality to
> httpd. As I've said, the current apachectl is over-complicated and the
> split between apachectl and httpd is confusing to some people. This
> change would clear that up.
Various distros out there are also choosing to not even include apachectl
and instead include their own init.d start/stop script which invariably is
less powerful and sometimes plain wrong. It would be nice to include some
of that logic in the binary itself to give the package maintainers less
bullets for their feet.
-Rasmus
RE: is httpd a valid way to start Apache?
Posted by Joshua Slive <jo...@slive.ca>.
On Thu, 16 May 2002, Ryan Bloom wrote:
> My own opinion is that we leave things exactly as they are today. If
> you are running the binary by hand, you are taking some responsibility
> for knowing what you are doing. That means having the environment
> variables setup correctly before you start.
>
> If you don't want that responsibility, use apachectl to run the server.
> Trying to solve this problem any other way just seems like we are asking
> for trouble.
I think that is exactly what this proposal is saying. But at the same
time it is cleaning up apachectl and adding some useful functionality to
httpd. As I've said, the current apachectl is over-complicated and the
split between apachectl and httpd is confusing to some people. This
change would clear that up.
JOshua.
RE: is httpd a valid way to start Apache?
Posted by Ryan Bloom <rb...@covalent.net>.
My own opinion is that we leave things exactly as they are today. If
you are running the binary by hand, you are taking some responsibility
for knowing what you are doing. That means having the environment
variables setup correctly before you start.
If you don't want that responsibility, use apachectl to run the server.
Trying to solve this problem any other way just seems like we are asking
for trouble.
Ryan
----------------------------------------------
Ryan Bloom rbb@covalent.net
645 Howard St. rbb@apache.org
San Francisco, CA
> -----Original Message-----
> From: trawick@rdu88-251-253.nc.rr.com [mailto:trawick@rdu88-251-
> 253.nc.rr.com] On Behalf Of Jeff Trawick
> Sent: Thursday, May 16, 2002 7:35 AM
> To: dev@httpd.apache.org
> Subject: Re: is httpd a valid way to start Apache?
>
> Who would be in favor of this, or parts thereof? I'm anxious to see a
> consensus so that steps toward the goal can be filled in over time.
>
> 1) httpd can't be used directly on Unix because of environment
> variable settings
>
> or "shouldn't", take your pick; I suspect that we're better off if
> httpd doesn't work anymore without some magic that people debugging
> it would know (maybe AP_DEBUG allows it to work directly)
>
> (Personally I'd prefer that httpd is the wrapper script, but I see
> one person definitely against that and one person definitely for
> apachectl being the wrapper script.)
>
> 2) httpd picks up the following function from apachectl:
>
> a) "apachectl start" -> "httpd -k start"
> (it is different from plain "httpd")
>
> b) "apachectl startssl" -> "httpd -DSSL -k start"
>
> c) "apachectl stop" -> "httpd -k stop"
>
> d) "apachectl restart" -> "httpd -k restart"
>
> e) "apachectl graceful" -> "httpd -k graceful"
>
> 3) apachectl becomes a simple wrapper script for httpd
>
> for the time being, most existing apachectl features will be mapped
> to their httpd equivalents
>
> a) "apachectl configtest" is mapped to the existing "httpd -t"
> invocation
>
> b) the apachectl commands listed above are translated into their
> new httpd equivalents, but generally we expect to pass the
> command-line through verbatim to httpd
>
> the following features are lost:
>
> a) "apachectl fullstatus"
>
> b) "apachectl status"
>
> c) "apachectl help"
>
> --
> Jeff Trawick | trawick@attglobal.net
> Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Andrew Ho <an...@tellme.com>.
Hello,
JM>Due to inheritance (export) of environment variables I usually start
JM>Apache after doing a:
JM># for i in `export | cut -f3 -d' '|cut -f1 -d'='`; do export -n $i; done
JM>Which cleans them all up nicely. One reason for doing this is so that
JM>phpinfo() shows all these nice Enviroment variables which could reveal
JM>one thing or the other of some persons setup. Maybe it's an idea to let
JM>the 'startup' program (may it be httpd or apachectl) to filter these
JM>environment variables too ? "they also take up memory", oooh ah :)
In our production environment we also include such functionality in our
apachectl script. For some variables (LD_LIBRARY_PATH for example) it is
essential to do this to make sure things will work, and for others (say,
PATH) it's just annoying to pick up different values depending on who
su'ed before starting a server.
However, I don't know that this should be in the default Apache
distribution. Often you'd want (or need) LD_LIBRARY_PATH or PATH or even
ORACLE_HOME to be inherited on purpose (hence PassEnv).
By the way, as someone who maintains Apache installations on many
platforms, I definitely like moving the pidfile checking out of apachectl.
I like the httpd -k idea to simplify apachectl.
Humbly,
Andrew
----------------------------------------------------------------------
Andrew Ho http://www.tellme.com/ andrew@tellme.com
Engineer info@tellme.com Voice 650-930-9062
Tellme Networks, Inc. 1-800-555-TELL Fax 650-930-9101
----------------------------------------------------------------------
RE: is httpd a valid way to start Apache?
Posted by Jeroen Massar <je...@unfix.org>.
Jeff Trawick wrote:
> Who would be in favor of this, or parts thereof? I'm anxious to see a
> consensus so that steps toward the goal can be filled in over time.
>
> 1) httpd can't be used directly on Unix because of environment
> variable settings
<SNIP>
Due to inheritance (export) of environment variables I usually start
Apache after doing a:
# for i in `export | cut -f3 -d' '|cut -f1 -d'='`; do export -n $i; done
Which cleans them all up nicely. One reason for doing this is so that
phpinfo() shows all these nice
Enviroment variables which could reveal one thing or the other of some
persons setup.
Maybe it's an idea to let the 'startup' program (may it be httpd or
apachectl) to filter these environment
variables too ? "they also take up memory", oooh ah :)
Greets,
Jeroen
Re: is httpd a valid way to start Apache?
Posted by Cliff Woolley <jw...@virginia.edu>.
On 16 May 2002, Jeff Trawick wrote:
> Step 3 is what removes the complexity from apachectl. Step 2 is what
> gives httpd the features required for step 3.
I guess I don't see:
start)
stop)
graceful)
restart)
httpd -k $1
;;
startssl)
httpd -DSSL -k start
;;
or whatever the syntax is for #2 alone as very complex. The complexity to
which I was referring was that right now apachectl has to know where the
PID file is, read the PID file, kill that pid, etc, etc.
I'll just say +0 to #3.
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Cliff Woolley <jw...@virginia.edu> writes:
> > 3) apachectl becomes a simple wrapper script for httpd
>
> You could do both 2 and 3 I suppose, but ... nah. #2 is enough to remove
> the complexity from apachectl.
Step 3 is what removes the complexity from apachectl. Step 2 is what
gives httpd the features required for step 3.
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Cliff Woolley <jw...@virginia.edu>.
On 16 May 2002, Jeff Trawick wrote:
> Who would be in favor of this, or parts thereof? I'm anxious to see a
> consensus so that steps toward the goal can be filled in over time.
>
> 1) httpd can't be used directly on Unix because of environment
> variable settings
Fine.
>
> or "shouldn't", take your pick; I suspect that we're better off if
> httpd doesn't work anymore without some magic that people debugging
> it would know (maybe AP_DEBUG allows it to work directly)
httpd should definitely not become the wrapper script. Leave httpd as the
binary and apachectl as the wrapper.
> 2) httpd picks up the following function from apachectl:
>
> a) "apachectl start" -> "httpd -k start"
> (it is different from plain "httpd")
Fine. More consistent with Win32.
> 3) apachectl becomes a simple wrapper script for httpd
You could do both 2 and 3 I suppose, but ... nah. #2 is enough to remove
the complexity from apachectl.
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Joshua Slive <jo...@slive.ca> writes:
> On 16 May 2002, Jeff Trawick wrote:
>
> > Who would be in favor of this, or parts thereof? I'm anxious to see a
> > consensus so that steps toward the goal can be filled in over time.
>
> +1 That looks good to me. Really, I have no particular problem with httpd
> being the wrapper script, but it would certainly cause more disruption to
> existing setups.
>
> A couple points of clarification:
>
> - I think I would recommend a "This is deprecated" message for the
> non -k options so that they don't need to be documented and supported
> indefinitely. Then we can make it clear that there is only one true set
> of command line options and they apply both to apachectl and httpd.
>
> - httpd -k restart and httpd -k graceful should do a config check before
> restarting and refuse to restart if the config is not valid.
agreed 100%
and "httpd -k start" will make sure that it isn't already running
(well, to the extent that apachectl does today)...
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Greg Ames <gr...@apache.org>.
Joshua Slive wrote:
> On 16 May 2002, Jeff Trawick wrote:
>
> > Who would be in favor of this, or parts thereof? I'm anxious to see a
> > consensus so that steps toward the goal can be filled in over time.
>
> +1 That looks good to me. Really, I have no particular problem with httpd
> being the wrapper script, but it would certainly cause more disruption to
> existing setups.
What Joshua said. This is going to be somewhat disruptive no matter which way
we go, including doing nothing.
Greg
Re: is httpd a valid way to start Apache?
Posted by Greg Ames <gr...@apache.org>.
Joshua Slive wrote:
>
> On 16 May 2002, Jeff Trawick wrote:
>
> > Who would be in favor of this, or parts thereof? I'm anxious to see a
> > consensus so that steps toward the goal can be filled in over time.
>
> +1 That looks good to me. Really, I have no particular problem with httpd
> being the wrapper script, but it would certainly cause more disruption to
> existing setups.
>
> A couple points of clarification:
>
> - I think I would recommend a "This is deprecated" message for the
> non -k options so that they don't need to be documented and supported
> indefinitely. Then we can make it clear that there is only one true set
> of command line options and they apply both to apachectl and httpd.
>
> - httpd -k restart and httpd -k graceful should do a config check before
> restarting and refuse to restart if the config is not valid.
>
> Joshua.
Re: is httpd a valid way to start Apache?
Posted by Joshua Slive <jo...@slive.ca>.
On 16 May 2002, Jeff Trawick wrote:
> Who would be in favor of this, or parts thereof? I'm anxious to see a
> consensus so that steps toward the goal can be filled in over time.
+1 That looks good to me. Really, I have no particular problem with httpd
being the wrapper script, but it would certainly cause more disruption to
existing setups.
A couple points of clarification:
- I think I would recommend a "This is deprecated" message for the
non -k options so that they don't need to be documented and supported
indefinitely. Then we can make it clear that there is only one true set
of command line options and they apply both to apachectl and httpd.
- httpd -k restart and httpd -k graceful should do a config check before
restarting and refuse to restart if the config is not valid.
Joshua.
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Who would be in favor of this, or parts thereof? I'm anxious to see a
consensus so that steps toward the goal can be filled in over time.
1) httpd can't be used directly on Unix because of environment
variable settings
or "shouldn't", take your pick; I suspect that we're better off if
httpd doesn't work anymore without some magic that people debugging
it would know (maybe AP_DEBUG allows it to work directly)
(Personally I'd prefer that httpd is the wrapper script, but I see
one person definitely against that and one person definitely for
apachectl being the wrapper script.)
2) httpd picks up the following function from apachectl:
a) "apachectl start" -> "httpd -k start"
(it is different from plain "httpd")
b) "apachectl startssl" -> "httpd -DSSL -k start"
c) "apachectl stop" -> "httpd -k stop"
d) "apachectl restart" -> "httpd -k restart"
e) "apachectl graceful" -> "httpd -k graceful"
3) apachectl becomes a simple wrapper script for httpd
for the time being, most existing apachectl features will be mapped
to their httpd equivalents
a) "apachectl configtest" is mapped to the existing "httpd -t"
invocation
b) the apachectl commands listed above are translated into their
new httpd equivalents, but generally we expect to pass the
command-line through verbatim to httpd
the following features are lost:
a) "apachectl fullstatus"
b) "apachectl status"
c) "apachectl help"
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
At 01:26 PM 5/15/2002, you wrote:
>On Wed, May 15, 2002 at 11:21:04AM -0700, Justin Erenkrantz wrote:
> > Is there any reason that httpd couldn't read/parse/source envvars
> > file itself? That sort of sucks, but perhaps that's a way to
> > get around this? -- justin
>
>Naw, you get dynamic linking failures as soon as you exec() the
>process.
Because it's not found... question; is it possible to link the .so's with a
path
of ./libfoo.so rather than /long/path/to/libfoo.so? Drop the libs in with the
binaries and away you go? Just a thought.
Bill
Re: is httpd a valid way to start Apache?
Posted by Justin Erenkrantz <je...@apache.org>.
On Wed, May 15, 2002 at 11:55:37AM -0700, Marc Slemko wrote:
> On Wed, 15 May 2002, Justin Erenkrantz wrote:
>
> > pkill -HUP httpd would all fail if the binary isn't called httpd.
>
> Erm... that isn't a bad thing.
I'm assuming you mean that I should *only* do:
ps -p `cat /where/my/pid/file/that/i/placed/in/my/conf/file/is`
kill -HUP `cat /where/my/pid/file/that/i/placed/in/my/conf/file/is`
There's more than one way to skin the poor cat. -- justin
Re: is httpd a valid way to start Apache?
Posted by Manoj Kasichainula <ma...@io.com>.
On Thu, May 16, 2002 at 10:16:56PM -0701, Jos Backus wrote:
> On Thu, May 16, 2002 at 07:27:46PM -0700, Manoj Kasichainula wrote:
> > I've (mostly) written replacements for supervise, setuidgid, and
> > tcpserver. They use Single Unix APIs, haven't been ported to APR, and
> > have no docs yet, but they are working for me.
> >
> > I imagine porting them to APR wouldn't be too painful, though they
> > wouldn't remain the svelte 4-8kB binaries they are today. :)
>
> Interesting. I'm not sure how much benefit there would be from using APR
> though.
Mainly portability to older Unixes that don't support some of the more
modern calls I used (or that break Single Unix in ways that Linux
doesn't). I guess autoconf, etc. would do the job as well.
> > Are people interested in this code?
>
> I for one would be interested in seeing this.
http://www.io.com/~manoj/file/mktool-0.0.7.tar.gz
I've only built it on my Linux boxes; I haven't even tried FreeBSD
yet, though I did try to avoid Linuxisms. The supervise replacement is
called "babysit" (so it wouldn't be confused with the djb tool it
tries to work as).
I'll hopefully clean it up a bit more, and maybe even test and doc it
more next week.
> I may be able to sell this to
> the FreeBSD people for inclusion in the base OS if the license allows it.
That won't be a problem. :)
> Do you also have equivalents to svc, svstat and svok?
I haven't written svc or svstat replacements yet; I've just used 'echo
-n "dx" > directory/control' in my scripts in the mean time. But they
will be easy. There is an svok workalike called "bsok" in the tarball
though.
Re: is httpd a valid way to start Apache?
Posted by Jos Backus <jo...@catnook.com>.
On Thu, May 16, 2002 at 07:27:46PM -0700, Manoj Kasichainula wrote:
> On Wed, May 15, 2002 at 12:49:46PM -0701, Jos Backus wrote:
> > It would be great to have a BSD-licensed version of
> > something like djb's daemontools.
>
> I've (mostly) written replacements for supervise, setuidgid, and
> tcpserver. They use Single Unix APIs, haven't been ported to APR, and
> have no docs yet, but they are working for me.
>
> I imagine porting them to APR wouldn't be too painful, though they
> wouldn't remain the svelte 4-8kB binaries they are today. :)
Interesting. I'm not sure how much benefit there would be from using APR
though.
> supervise /var/supervise/producer | supervise /var/supervise/consumer
Thanks for the nice explanations.
> In my supervise replacement, I use a technique picked up from the
> Awesome Might that is Dean (and which was used in some of the early
> MPMs). Create a pipe, and keep both ends of it. Create a signal
> handler for SIGCHLD that writes to the pipe. And, in the select() call
> you reference, wait on both the sigchld pipe (a.k.a. pipe_of_death)
> and the named fifo.
This sounds like the self-pipe trick (see http://cr.yp.to/docs/selfpipe.html).
Neat.
> Are people interested in this code?
I for one would be interested in seeing this. I may be able to sell this to
the FreeBSD people for inclusion in the base OS if the license allows it.
Do you also have equivalents to svc, svstat and svok?
--
Jos Backus _/ _/_/_/ Santa Clara, CA
_/ _/ _/
_/ _/_/_/
_/ _/ _/ _/
jos@catnook.com _/_/ _/_/_/ use Std::Disclaimer;
Re: is httpd a valid way to start Apache?
Posted by Manoj Kasichainula <ma...@io.com>.
On Wed, May 15, 2002 at 12:49:46PM -0701, Jos Backus wrote:
> What about moving into the other direction and moving the process management
> portion into a separate set of tools so it can be used with other daemons
> besides httpd?
I've pondered writing something like this, but then I also ponder the
opposite. Why not build the supervise support into httpd itself. httpd
would listen on a unix-domain socket or FIFO in the filesystem, and
all graceful signals would be replaced with writes to this fifo. This
eliminates the crufty signal layer between the supervise replacement
and httpd, completely hides the difference between Unix and Windows
(using the -k syntax) and potentially allows for more interesting
status reporting from the command-line. Probably not viable for 2.0,
though.
> It would be great to have a BSD-licensed version of
> something like djb's daemontools.
I've (mostly) written replacements for supervise, setuidgid, and
tcpserver. They use Single Unix APIs, haven't been ported to APR, and
have no docs yet, but they are working for me.
I imagine porting them to APR wouldn't be too painful, though they
wouldn't remain the svelte 4-8kB binaries they are today. :)
On Wed, May 15, 2002 at 07:15:47PM -0701, Jos Backus wrote:
> Seriously, a decent process controller that would allow starting, stopping and
> sending various signals to a command that runs as its child (i.e. duplicating
> supervise's functionality) should not be too hard to implement.
Nope, it wasn't. Assuming mine works, anyway.
> It's the
> fiddling with the pipes between two supervise's (the main one and the log/
> one) that seems tricky to me.
I don't think this is a problem, because of the way Unix mangles the
file descriptors. If you do something like:
supervise /var/supervise/producer | supervise /var/supervise/consumer
then, your shell will arrange the fd's so that stdout for producer
points to a pipe to stdin for consumer. Then, when the supervise
processes fork-exec their children, the kids inherit the 2 ends of the
pipe, and just use them as if the pipe were set up just for them. If
the producer dies, there's no problem, because the parent supervise
still keeps an fd for that end of the pipe and passes it on to the
next generation of the producer. Same goes for the consumer. It might
be nice to use a technique like this to take the code for reliable
piped logging out of httpd and make it more generally useful.
> And I'd not sure how you'd wait() for the child
> while still being able to select() on a name pipe in order to read control
> messages sent by svc.
In my supervise replacement, I use a technique picked up from the
Awesome Might that is Dean (and which was used in some of the early
MPMs). Create a pipe, and keep both ends of it. Create a signal
handler for SIGCHLD that writes to the pipe. And, in the select() call
you reference, wait on both the sigchld pipe (a.k.a. pipe_of_death)
and the named fifo.
Are people interested in this code?
Re: is httpd a valid way to start Apache?
Posted by Jos Backus <jo...@catnook.com>.
I wrote:
> It would be great to have a BSD-licensed version of something like djb's
> daemontools.
> That would be totally fine/cool. Are you volunteering? :-)
I would if I was even half as decent a programmer as some of you :-)
Seriously, a decent process controller that would allow starting, stopping and
sending various signals to a command that runs as its child (i.e. duplicating
supervise's functionality) should not be too hard to implement. It's the
fiddling with the pipes between two supervise's (the main one and the log/
one) that seems tricky to me. And I'd not sure how you'd wait() for the child
while still being able to select() on a name pipe in order to read control
messages sent by svc. But those details can be worked out, and yes, I'd
certainly be willing to lend a hand. The main point is that if such a set of
tools were to be made available with a friendly license, would people such as
the Apache developers promote them and use them? UNIX needs a standard
cross-platform process monitoring solution so individual daemons can stop
reinventing the wheel, and the standard control interface will make life
easier for sysadmins.
--
Jos Backus _/ _/_/_/ Santa Clara, CA
_/ _/ _/
_/ _/_/_/
_/ _/ _/ _/
jos@catnook.com _/_/ _/_/_/ use Std::Disclaimer;
Re: is httpd a valid way to start Apache?
Posted by Greg Stein <gs...@lyra.org>.
On Wed, May 15, 2002 at 12:49:46PM -0701, Jos Backus wrote:
> What about moving into the other direction and moving the process management
> portion into a separate set of tools so it can be used with other daemons
> besides httpd? This would also improve portability of the mechanism: no
> worries about pid file races and staleness, differing versions of
> ps/kill/killall/etc. It would be great to have a BSD-licensed version of
> something like djb's daemontools.
That would be totally fine/cool. Are you volunteering? :-)
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
Re: is httpd a valid way to start Apache?
Posted by Jos Backus <jo...@catnook.com>.
What about moving into the other direction and moving the process management
portion into a separate set of tools so it can be used with other daemons
besides httpd? This would also improve portability of the mechanism: no
worries about pid file races and staleness, differing versions of
ps/kill/killall/etc. It would be great to have a BSD-licensed version of
something like djb's daemontools.
--
Jos Backus _/ _/_/_/ Santa Clara, CA
_/ _/ _/
_/ _/_/_/
_/ _/ _/ _/
jos@catnook.com _/_/ _/_/_/ use Std::Disclaimer;
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Justin Erenkrantz <je...@apache.org> writes:
> > How many problems does it cause? Having to give gdb a different name
> > is a problem (minor), but how many others are there?
>
> ps and other process utilities won't filter on httpd. ps -C httpd,
> pkill -HUP httpd would all fail if the binary isn't called httpd.
gotcha (but I'm used to lt-httpd by now anyway :) )
> Perhaps we could so some sort of transformation for binbuilds that
> only adds the httpd.bin to those distributions. -- justin
That is doable, and covers the most serious problem, but on platforms
where other environment variables are needed it won't solve the
remaining problems, so apachectl will still need to source envvars.
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Marc Slemko <ma...@znep.com>.
On Wed, 15 May 2002, Justin Erenkrantz wrote:
> pkill -HUP httpd would all fail if the binary isn't called httpd.
Erm... that isn't a bad thing.
Re: is httpd a valid way to start Apache?
Posted by Joshua Slive <jo...@slive.ca>.
On 15 May 2002, Jeff Trawick wrote:
> Joshua Slive <jo...@slive.ca> writes:
>
> > What about
> >
> > - Incorporate the restart stuff into httpd
> > - Make apachectl a simple shell script that just sets up the environment
> > and calls httpd
> > - Encourage everyone to use apachectl
>
> Would the syntax for start/restart/configtest be exactly the same as
> current apachectl, or would it be changed? (I think it would be
> changed to use "-k keyword" for apachectl options that don't
> correspond directly to existing httpd options.)
>
It would be very nice if "httpd graceful" could be made a synonym
for "httpd -k graceful", but I don't think the current getopt
stuff will handle that. Otherwise, I would favor an apachectl
that looks something like
#!/bin/sh
# pick up any necessary environment variables
if test -f @exp_bindir@/envvars; then
. @exp_bindir@/envvars
fi
case $@ in
start)
shift 1
;;
stop)
COMMAND=-k stop
shift 1
;;
graceful)
COMMAND=-k graceful
shift 1
;;
# etc...
httpd $COMMAND $@
(Please excuse the horrible shell script, but you get the point.)
Possibly with a "this is deprecated please use -k" message
in the appropriate places.
This is not perfectly compatible with apachectl, but it is pretty
close, and very much simpler.
Joshua.
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Joshua Slive <jo...@slive.ca> writes:
> What about
>
> - Incorporate the restart stuff into httpd
> - Make apachectl a simple shell script that just sets up the environment
> and calls httpd
> - Encourage everyone to use apachectl
Would the syntax for start/restart/configtest be exactly the same as
current apachectl, or would it be changed? (I think it would be
changed to use "-k keyword" for apachectl options that don't
correspond directly to existing httpd options.)
If the syntax changes, then maybe
1) apachectl remains for a while but is deprecated; it shrinks to
something that translates between existing apachectl syntax and the
new httpd options
2) a new script called apache sets up the environment and calls httpd;
its command-line syntax is exactly the same as httpd
(oh, Apache isn't just the Apache httpd server? I guess this name
isn't so good)
3) httpd is the binary which we don't want anybody to run directly
unless they know what they are doing
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Joshua Slive <jo...@slive.ca>.
On Wed, 15 May 2002, Justin Erenkrantz wrote:
> > How many problems does it cause? Having to give gdb a different name
> > is a problem (minor), but how many others are there?
>
> ps and other process utilities won't filter on httpd. ps -C httpd,
> pkill -HUP httpd would all fail if the binary isn't called httpd.
>
> Perhaps we could so some sort of transformation for binbuilds that
> only adds the httpd.bin to those distributions. -- justin
Hmmm... don't like inconsistency in distributions.
What about
- Incorporate the restart stuff into httpd
- Make apachectl a simple shell script that just sets up the environment
and calls httpd
- Encourage everyone to use apachectl
This makes it much clearer that there is a binary (httpd) that needs to be
used for ps/pkill/gdb, and a seperate control program.
It also solves the problem that the current apachectl is over-complicated
and confusing to some users. Problems with current apachectl
include:
- there are two sets of command line options: httpd and apachectl, and it
is difficult to tell which ones go where
- apachectl needs to know the location of the pid file, and users don't
understand they need to edit this when they change their configuration
Joshua.
Re: is httpd a valid way to start Apache?
Posted by Justin Erenkrantz <je...@apache.org>.
On Wed, May 15, 2002 at 02:44:37PM -0400, Jeff Trawick wrote:
> Wouldn't we have to do a full link-edit on the target machine to make
> that work portably? One of the nice things about binbuilds is that we
> don't give a rip whether the user has a working ld (just an example).
> We don't want to care whether or not somebody has working binutils.
I'm not sure what's involved on each platform. I think I remember
reading on some platforms, libtool could be tricky and do it without
needing a full recompile or relink (it just modifies the binary
itself). I'm not sure if that's even possible or not (doesn't seem
so). But, yeah, I could see it causing problems on esoteric
platforms. Perhaps libtool 1.4.x can do this better than 1.3?
> How many problems does it cause? Having to give gdb a different name
> is a problem (minor), but how many others are there?
ps and other process utilities won't filter on httpd. ps -C httpd,
pkill -HUP httpd would all fail if the binary isn't called httpd.
Perhaps we could so some sort of transformation for binbuilds that
only adds the httpd.bin to those distributions. -- justin
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Justin Erenkrantz <je...@apache.org> writes:
> On Wed, May 15, 2002 at 11:26:42AM -0700, Aaron Bannert wrote:
> > Naw, you get dynamic linking failures as soon as you exec() the
> > process.
>
> Yeah, that's true. Moving the start/restart logic into httpd still
> seems worthy though.
yeah
> I know that libtool can (supposedly) relink binaries to change the
> link paths. Could that be something we could do?
Wouldn't we have to do a full link-edit on the target machine to make
that work portably? One of the nice things about binbuilds is that we
don't give a rip whether the user has a working ld (just an example).
We don't want to care whether or not somebody has working binutils.
> Using a shell script is going to cause more problems than it helps -
> I like being able to do:
>
> gdb /usr/local/apache2/bin/httpd core.file
How many problems does it cause? Having to give gdb a different name
is a problem (minor), but how many others are there?
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Greg Ames <gr...@apache.org>.
Justin Erenkrantz wrote:
> I know that libtool can (supposedly) relink binaries to change the
> link paths. Could that be something we could do?
I don't think that libtool has any magic that will produce a major breakthrough
here. I played around with it back when binbuild was totally non-functional,
and was underwhelmed.
Greg
Re: is httpd a valid way to start Apache?
Posted by Justin Erenkrantz <je...@apache.org>.
On Wed, May 15, 2002 at 11:26:42AM -0700, Aaron Bannert wrote:
> Naw, you get dynamic linking failures as soon as you exec() the
> process.
Yeah, that's true. Moving the start/restart logic into httpd still
seems worthy though.
I know that libtool can (supposedly) relink binaries to change the
link paths. Could that be something we could do?
Using a shell script is going to cause more problems than it helps -
I like being able to do:
gdb /usr/local/apache2/bin/httpd core.file
Mozilla went down this route of a fake binary - I'm not sure I like
that approach. -- justin
Re: is httpd a valid way to start Apache?
Posted by Aaron Bannert <aa...@clove.org>.
On Wed, May 15, 2002 at 11:21:04AM -0700, Justin Erenkrantz wrote:
> Is there any reason that httpd couldn't read/parse/source envvars
> file itself? That sort of sucks, but perhaps that's a way to
> get around this? -- justin
Naw, you get dynamic linking failures as soon as you exec() the
process.
-aaron
Re: is httpd a valid way to start Apache?
Posted by Jeff Trawick <tr...@attglobal.net>.
Justin Erenkrantz <je...@apache.org> writes:
> On Wed, May 15, 2002 at 02:17:55PM -0400, Jeff Trawick wrote:
> > If httpd is to be left as-is on Unix (i.e., maybe it works maybe it
> > doesn't), then perhaps apachectl needs to be pushed as the way to
> > access all httpd functionality.
>
> One thing that I seem to recall is that we've talked about taking
> the apachectl functionality and moving it into httpd (the C
> program that is - not renaming apachectl as httpd). So, you can do:
>
> httpd -k start
> httpd -k stop
> httpd -k graceful
>
> (ISTR that -k was the option in Win32 and for various reasons, we
> wanted to use that to indicate the command to run.)
>
> Is there any reason that httpd couldn't read/parse/source envvars
> file itself? That sort of sucks, but perhaps that's a way to
> get around this? -- justin
It can, and that would fix certain things, like the OS/390 and AIX
example of needing to tune the memory use to allow lots of threads.
(Though I wonder how it would find the right copy of envvars in a
binbuild situation.)
Unfortunately, it wouldn't fix the inability of a binbuild httpd from
finding its libraries. A wrapper script seems to cover both types of
problems.
--
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...
Re: is httpd a valid way to start Apache?
Posted by Justin Erenkrantz <je...@apache.org>.
On Wed, May 15, 2002 at 02:17:55PM -0400, Jeff Trawick wrote:
> If httpd is to be left as-is on Unix (i.e., maybe it works maybe it
> doesn't), then perhaps apachectl needs to be pushed as the way to
> access all httpd functionality.
One thing that I seem to recall is that we've talked about taking
the apachectl functionality and moving it into httpd (the C
program that is - not renaming apachectl as httpd). So, you can do:
httpd -k start
httpd -k stop
httpd -k graceful
(ISTR that -k was the option in Win32 and for various reasons, we
wanted to use that to indicate the command to run.)
Is there any reason that httpd couldn't read/parse/source envvars
file itself? That sort of sucks, but perhaps that's a way to
get around this? -- justin