apachectl 2010-12-20

[Lewis G Rosenthal]

Steve -

This is more useful to me than the earlier builds, though it will still
require quite a bit of tweaking. The hardcoded hostnames were easy to
suss out, but my own process killers don't want to actually kill with
what's being passed to them, and startup doesn't want to happen. Still,
I should be able to make things work as desired.

Thanks!

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE, CWTS
Rosenthal& Rosenthal, LLC www.2rosenthals.com
Need a managed Wi-Fi hotspot? www.hautspot.com
visit my IT blog www.2rosenthals.net/wordpress
please do not add my address to any non-bcc mass mailings
-------------------------------------------------------------

[Steven Levine]

In <4EE18002...@2rosenthals.com>, on 12/08/11
at 10:26 PM, Lewis G Rosenthal <lgros...@2rosenthals.com> said:

Hi,

>require quite a bit of tweaking.

It should not be that much tweaking.

>The hardcoded hostnames were easy to suss out,

The error message does tell you which line to look at. I will think about
adding some logic that checks if the current directory is a usable apache
directory.

>but my own process killers don't want to actually kill with
>what's being passed to them,

Which process killers? Do they support unix style signals? I have no
reason to support killers like pkill or go for exactly the reason Massimo
mentioned.

>and startup doesn't want to happen.

Use the -f option so you can see the failure message(s).

Steven

--
----------------------------------------------------------------------
"Steven Levine" <ste...@earthlink.net> eCS/Warp/DIY etc.
www.scoug.com www.ecomstation.com
----------------------------------------------------------------------

[Lewis G Rosenthal]

On 12/09/11 11:24 am, Steven Levine thus wrote :

> In<4EE18002...@2rosenthals.com>, on 12/08/11
> at 10:26 PM, Lewis G Rosenthal<lgros...@2rosenthals.com> said:
>
> Hi,
>
>> require quite a bit of tweaking.
> It should not be that much tweaking.
>

Paths, process killer. I haven't looked through all the code, yet.

>> The hardcoded hostnames were easy to suss out,
> The error message does tell you which line to look at. I will think about
> adding some logic that checks if the current directory is a usable apache
> directory.
>

:-) Indeed! The error message is quite helpful.

>> but my own process killers don't want to actually kill with
>> what's being passed to them,
> Which process killers? Do they support unix style signals? I have no
> reason to support killers like pkill or go for exactly the reason Massimo
> mentioned.
>

The first in my path is kill (coreutils), which did *not* kill the
process. I agree with not supporting pkill or go, but what about killall
or kill9? (I use killall on Linux; I don't think I have it on any OS/2
boxes.) Also, while kill is in the path, I had to define it in the
script. This isn't a complaint; it's an observation.

>> and startup doesn't want to happen.
> Use the -f option so you can see the failure message(s).
>

Thanks. will do.
:-)

[Massimo]

Il 10/12/2011 1.32, Lewis G Rosenthal ha scritto:
> The first in my path is kill (coreutils), which did *not* kill the process. I agree with not
> supporting pkill or go,

i pray you, i can't live without go.exe :)

massimo s.

[Steven Levine]

In <4EE2A8A0...@2rosenthals.com>, on 12/09/11
at 07:32 PM, Lewis G Rosenthal <lgros...@2rosenthals.com> said:

Hey,

>The first in my path is kill (coreutils), which did *not* kill the
>process.

Perhaps it's broken? The --help says it's supposed to understand all the
standard signal names so it should be compatible with apache_kill.

>but what about killall
>or kill9?

How are they better than apache_kill for our purposes?

>Also, while kill is in the path, I had to define it in the
>script.

That's by design. None of the systems that I use apachectl require that I
use kill. Given your recent experiences with it, I'm not so sure I'm
missing anything. :-) FWIW, there are lots of broken process killers out
there.

[Steven Levine]

In <4EE2C40D...@ecomstation.it>, on 12/10/11
at 03:29 AM, Massimo <m...@ecomstation.it> said:

Hi,

>i pray you, i can't live without go.exe :)

No to worry. I will not be taking your copy away from you. :-)

[Lewis G Rosenthal]

On 12/10/2011 01:03 AM, Steven Levine wrote:
> In<4EE2A8A0...@2rosenthals.com>, on 12/09/11
> at 07:32 PM, Lewis G Rosenthal<lgros...@2rosenthals.com> said:
>
> Hey,
>
>> The first in my path is kill (coreutils), which did *not* kill the
>> process.
> Perhaps it's broken? The --help says it's supposed to understand all the
> standard signal names so it should be compatible with apache_kill.

Could be. I actually have a couple copies of kill.exe (one in \bin and
another in \usr\sbin). I'll diff them and see.

>> but what about killall
>> or kill9?
> How are they better than apache_kill for our purposes?

I suppose they're not. They're just the ones I typically use by
convention. On *nix, I like killall, because it takes a name as a param
vs just a PID. However, that can lead to some ambiguity (on my NetWare
box, here, I'm running four separate instances of Apache 2.2 in
different address spaces, listening on different IPs - all SSL; I need
to be a bit more particular as to which Apache2 process I want to
kill!). Obviously, one could run the same type of setup on any other OS
(well, perhaps not in completely separate address spaces, but surely,
more than one instance of httpd.exe, as long as the listeners didn't
conflict).

>> Also, while kill is in the path, I had to define it in the
>> script.
> That's by design. None of the systems that I use apachectl require that I
> use kill. Given your recent experiences with it, I'm not so sure I'm
> missing anything. :-) FWIW, there are lots of broken process killers out
> there.

No, I suppose not. My design philosophy leans more in the direction of
having the ability to set these variables in the environment or in a
config file, vs having to modify each copy of the script on each
machine, that's all. No big deal. Once I get the paths set, though, and
make sure the process killer really *is* working (CommuniGate's kill.exe
is hopelessly broken, and falls into the category you mention, above), I
should be fine with this. Thanks again for providing it!

Cheers, buddy. Have a good weekend.

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE, CWTS
Rosenthal& Rosenthal, LLC www.2rosenthals.com
Need a managed Wi-Fi hotspot? www.hautspot.com
visit my IT blog www.2rosenthals.net/wordpress

-------------------------------------------------------------

[Steven Levine]

In <4EE3053B...@2rosenthals.com>, on 12/10/11
at 02:07 AM, Lewis G Rosenthal <lgros...@2rosenthals.com> said:

Hi,

>convention. On *nix, I like killall, because it takes a name as a param
>vs just a PID.

A couple of the eCS/OS2 kills can do this, pkill for example.

>No, I suppose not. My design philosophy leans more in the direction of
>having the ability to set these variables in the environment or in a
>config file, vs having to modify each copy of the script on each
>machine, that's all.

If I had planned to make this publically available, the script would have
generic and configurable. Instead, it started out as something for
internal use and then ended up being used on some of Dan's servers.

>Thanks again for providing it!

You're welcome. Have fun.