On Wed, 28 Jun 2006, Lars Marowsky-Bree wrote:
> On 2006-06-28T17:32:32, David Lee <[EMAIL PROTECTED]> wrote:
>
> > I was wondering whether the protocol is described somewhere (in similar
> > manner to RFCs describing SMTP, TCP, etc.). Alternatively, whether the
> > expected (designed) internal operation of lrmd was documented.
>
> Well, the protocol is not meant to be re-implemented anywhere, so the
> documentation seems to be in the code itself.
>
> A slightly more formal description would be cool, but I don't think that
> rates very high on our list.
>
> > The hanging "lrmadmin" was from "STONITHDBasicSanityCheck" where it does
> > the "lrmadmin -E s1 start 0 0 0".
> >
> > Note that on a different subthread of this topic, Alan has suggested that
> > I open a bugzilla report. (And Matt Soffen, another heartbeat developer
> > using non-Linux, has also just reported a problem.)
> >
> > So a little documentation about how it is _designed_ to work would help
> > folk like Matt and me chase problems.
>
> This looks more as if lrmd/lrmadmin can't even properly communicate.
Thanks, Lars. Some communication does happen:
shiel# /users/dcl0tdl/heartbeat/2.0/sparc.sunos5.9.csw/lrm/lrmd/.libs/lrmd
-r &
[1] 23357
shiel# /opt/LXHAhb/lib/heartbeat/lrmadmin -L
Currently no resource is managed by LRM.
shiel# /opt/LXHAhb/lib/heartbeat/lrmadmin -A s1 stonith null NULL
hostlist=shiel
Succeeded in adding this resource.
shiel# /opt/LXHAhb/lib/heartbeat/lrmadmin -L
Resource ID:s1
Resource agency class:stonith
Resource agency type:null
Resource agency provider:default
Resource agency parameters:hostlist=shiel
shiel# /opt/LXHAhb/lib/heartbeat/lrmadmin -E s1 start 0 0 0
waiting for calling result from the lrmd.
^Cshiel#
It is that last ("... -E ...") item that hangs (until "Ctrl-C" out of it).
> My uneducated guess would be that you don't need the lrmd documented,
> but the IPC layer we use, and then make sure that the streams
> implementation behaves exactly the same. Perhaps even sockets (when not
> using streams) behave differently on different flavors of Unix, in
> particular wrt non-blocking use.
The streams-specific code I wrote is remarkably small (fragments hidden in
the depths of "ipcsocket.c", revs 1.169 to 1.172). But I can well believe
that it might carry a lurking bug or two. (There is at least one
untidiness in the opening/closing behaviour.)
Seems like I need to open that bugzilla...
--
: David Lee I.T. Service :
: Senior Systems Programmer Computer Centre :
: Durham University :
: http://www.dur.ac.uk/t.d.lee/ South Road :
: Durham DH1 3LE :
: Phone: +44 191 334 2752 U.K. :
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
