[prev in list] [next in list] [prev in thread] [next in thread]
List: linux-ha-dev
Subject: Re: [Linux-ha-dev] [PATCH]A change of the external/STONITH log in
From: <renayama19661014 () ybb ! ne ! jp>
Date: 2009-12-18 0:23:20
Message-ID: 20091218002320.65319.qmail () web10403 ! mail ! ogk ! yahoo ! co ! jp
[Download RAW message or body]
Hi Dejan,
All right.
Thank you for a change to the development version of the patch which I offered.
Best Regards,
Hideo Yamauchi.
--- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> Hi Hideo-san,
>
> On Thu, Dec 17, 2009 at 09:33:37AM +0900, renayama19661014@ybb.ne.jp wrote:
> > Hi Dejan,
> >
> > > Well, didn't feel like doing it now, but that may be necessary. I've
> > > already got a patch, but then anything a plugin spits on the stdin
> > > (when doing power requests and status) goes to log at the info level
> > > and there's no way around it. For instance, that makes it impossible
> > > for plugins to log at the debug level.
> >
> > OK.
> > I understand.
> >
> > > I think we should revisit the whole external plugins logging and
> > > simply make them log directly through ha_log.
> >
> > The output by ha_log is surely necessary.
> > However, a revision to use "echo" for by a recent change enters.
> >
> > We must return this change again.
> > Is there any problem in our using ha_log?
>
> There was an issue with circular dependency because ha_log is
> defined in a file which belongs to resource agents. Anyway, we
> have to provide it. But we also need something for the python
> plugins. Perhaps a wrapper shell script would be better. I'll
> think about it. Unfortunately, so far all possibilities look
> pretty ugly.
>
> Cheers,
>
> Dejan
>
> > Best Regards,
> > Hideo Yamauchi.
> >
> > --- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> >
> > > Hi Hideo-san,
> > >
> > > On Wed, Dec 16, 2009 at 11:20:44AM +0900, renayama19661014@ybb.ne.jp wrote:
> > > > Hi Dejan,
> > > >
> > > > All right.
> > > >
> > > > >but I'll do something about it. We can track the
> > > > > issue in bugzilla 1972.
> > > >
> > > > Are you to revise it by a policy like lrmd?
> > >
> > > Well, didn't feel like doing it now, but that may be necessary. I've
> > > already got a patch, but then anything a plugin spits on the stdin
> > > (when doing power requests and status) goes to log at the info level
> > > and there's no way around it. For instance, that makes it impossible
> > > for plugins to log at the debug level.
> > >
> > > I think we should revisit the whole external plugins logging and
> > > simply make them log directly through ha_log.
> > >
> > > Cheers,
> > >
> > > Dejan
> > >
> > > > If I can assist your work, please mail to me.
> > > >
> > > > Best Regards,
> > > > Hideo Yamauchi.
> > > >
> > > > --- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> > > >
> > > > > Hi Hideo-san,
> > > > >
> > > > > On Tue, Dec 15, 2009 at 09:33:08AM +0900, renayama19661014@ybb.ne.jp wrote:
> > > > > > Hi Dejan,
> > > > > >
> > > > > > > The change is not difficult, but then we would need to change all
> > > > > > > external plugins too. It should've been done so in the first place.
> > > > > >
> > > > > > I understood it.
> > > > > > Will you adopt this patch till it is changed for processing same as lrmd?
> > > > > >
> > > > > > I attach a new patch.
> > > > >
> > > > > Thanks for the patch. It's still not clear to me what was your
> > > > > intention in the last part of the patch, which I quoted in the
> > > > > first reply, but I'll do something about it. We can track the
> > > > > issue in bugzilla 1972.
> > > > >
> > > > > Cheers,
> > > > >
> > > > > Dejan
> > > > >
> > > > > > Best Regards,
> > > > > > Hideo Yamauchi.
> > > > > >
> > > > > >
> > > > > > --- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> > > > > >
> > > > > > > Hi Hideo-san,
> > > > > > >
> > > > > > > On Mon, Dec 14, 2009 at 09:46:33AM +0900, renayama19661014@ybb.ne.jp wrote:
> > > > > > > > Hi Dejan,
> > > > > > > >
> > > > > > > > > It would also be good to rename "isbufferd" to say "data_call".
> > > > > > > > OK.
> > > > > > > >
> > > > > > > > > That way we state that what is expected from the plugin is data
> > > > > > > > > on stdin (getinfo, get_confignames, hostlist). With the other
> > > > > > > > > calls (status and reset_req) we expect informational or error
> > > > > > > > > messages.
> > > > > > > > I think it to change it to structure such as lrmd without using popen.
> > > > > > > > (lrmd : fork + exec + pipe(stdout,stdin))
> > > > > > > >
> > > > > > > > Is it a difficult change to make same as lrmd?
> > > > > > >
> > > > > > > The change is not difficult, but then we would need to change all
> > > > > > > external plugins too. It should've been done so in the first place.
> > > > > > >
> > > > > > > Thanks,
> > > > > > >
> > > > > > > Dejan
> > > > > > >
> > > > > > > > Best Regards,
> > > > > > > > Hideo Ymamauchi.
> > > > > > > >
> > > > > > > > --- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> > > > > > > >
> > > > > > > > > Hi Hideo-san,
> > > > > > > > >
> > > > > > > > > On Fri, Dec 11, 2009 at 09:50:13AM +0900, renayama19661014@ybb.ne.jp wrote:
> > > > > > > > > > Hi Dejan,
> > > > > > > > > >
> > > > > > > > > > Sorry....
> > > > > > > > > > Meanings of your comment cannot understand it well.
> > > > > > > > > > Please teach contents of the comment a little more intelligibly.
> > > > > > > > >
> > > > > > > > > Please ignore what I wrote. external_run_cmd uses popen() and the
> > > > > > > > > plugins use stdout for informational messages too. Not the best
> > > > > > > > > way, but we can live with that.
> > > > > > > > >
> > > > > > > > > I get the meaning of your patch, but still don't understand the
> > > > > > > > > part I quoted in the first reply.
> > > > > > > > >
> > > > > > > > > It would also be good to rename "isbufferd" to say "data_call".
> > > > > > > > > That way we state that what is expected from the plugin is data
> > > > > > > > > on stdin (getinfo, get_confignames, hostlist). With the other
> > > > > > > > > calls (status and reset_req) we expect informational or error
> > > > > > > > > messages.
> > > > > > > > >
> > > > > > > > > Cheers,
> > > > > > > > >
> > > > > > > > > Dejan
> > > > > > > > >
> > > > > > > > > > Best Regards,
> > > > > > > > > > Hideo Yamauchi.
> > > > > > > > > >
> > > > > > > > > > --- Dejan Muhamedagic <dejanmm@fastmail.fm> wrote:
> > > > > > > > > >
> > > > > > > > > > > Hi Hideo-san,
> > > > > > > > > > >
> > > > > > > > > > > On Thu, Dec 10, 2009 at 05:34:56PM +0900, renayama19661014@ybb.ne.jp wrote:
> > > > > > > > > > > > Hi,
> > > > > > > > > > > >
> > > > > > > > > > > > We are troubled with log in reset of stonith.
> > > > > > > > > > > >
> > > > > > > > > > > > Log is not output till practice is completed in the reset movement of
> stonith.
> > > > > > > > > > >
> > > > > > > > > > > I see your point.
> > > > > > > > > > >
> > > > > > > > > > > > The stonith module which is extenal which we made gives log during reset
> > > movement
> > > > > and
> > > > > > > > > chases
> > > > > > > > > > > movement.
> > > > > > > > > > > > For us, the chase of the reset movement is very useful.
> > > > > > > > > > > >
> > > > > > > > > > > > We think that the whole reset movement should be over the log at any time.
> > > > > > > > > > >
> > > > > > > > > > > The stdout/stderr should be split with stderr going immediately
> > > > > > > > > > > to the log and the content of stdout returned to the caller. Any
> > > > > > > > > > > chance to implement it that way.
> > > > > > > > > > >
> > > > > > > > > > > > I created a patch for external.c.
> > > > > > > > > > > > Please teach it if there is the revision method that, besides, is good.
> > > > > > > > > > >
> > > > > > > > > > > After taking a quick look, I find this part a bit strange:
> > > > > > > > > > >
> > > > > > > > > > > + if (fgets(buff, BUFF_LEN, file)) {
> > > > > > > > > > > + LOG(PIL_INFO, "%s: '%s' output: %s", __FUNCTION__, cmd, buff);
> > > > > > > > > > > + if (Debug && data) {
> > > > > > > > > > > + LOG(PIL_DEBUG, "%s: '%s' output: %s", __FUNCTION__, cmd, buff);
> > > > > > > > > > > + }
> > > > > > > > > > > + tosleep = FALSE;
> > > > > > > > > > > + }
> > > > > > > > > > >
> > > > > > > > > > > Cheers,
> > > > > > > > > > >
> > > > > > > > > > > Dejan
> > > > > > > > > > >
>
=== $B0J2<$N%a%C%;!<%8$O>JN,$5$l$^$7$?(B ===
_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic