y.
2. Editing of '/etc/passwd' and 'useradd'-like operations are not done by
'make install', and rightly so.
3. Editing of (etc.) is done by package-maintenance procedures, and
heartbeat supports this for Debian, RPM, pkg. (And BSD also, I think.)
4. Fo
of the hacluster and haclient user/group.
Now in place on the development version.
If you are really keen, see:
http://hg.linux-ha.org/dev/
The two changes I made are:
http://hg.linux-ha.org/dev/rev/3f4236a16927
http://hg.linux-ha.org/dev/rev/f4c534d86e55
Hope that is OK!
--
:
of platforms. I think I can
> recall somebody having a solaris/linux cluster.
(Drifting off-topic a little...)
To confirm: I have run a 1.2.x Solaris/Linux pair in the past. Never
really pushed it hard, and only used it for simple IPaddr failover, but it
worked that
d
probably be in one of the heartbeat runtime locations rather than /tmp.
If someone can identify the best location, then the autoconfiscation of
your code at the appropriate time will be able to handle this.
Sorry for that list! Please don't take it too nega
On Thu, 19 Apr 2007, Xinwei Hu wrote:
> [David Lee had earlier written:]
> >
> > 5. "ping -q -c 1 $ping_host". The options for "ping" are notoriously
> > variable from system to system. Keep it simple. (For example my system
> > doesn't have
On Fri, 20 Apr 2007, Alan Robertson wrote:
> David Lee wrote:
> > On Thu, 19 Apr 2007, Xinwei Hu wrote:
> >
> >> [David Lee had earlier written:]
> >>> 5. "ping -q -c 1 $ping_host". The options for "ping" are notoriously
> >>&
ubnets (as
in simple routing) where using the same MAC isn't a problem. (Concepts of
"ethernet bonding", "link aggregation" (etc.) of multiple interfaces are
relatively new and would require a little extra configuration to achieve
on default Sun hardware.)
Hope that helps.
On Fri, 27 Apr 2007, Alan Robertson wrote:
> David Lee wrote:
> >
> > The principle of calling the system "ping" is clean and simple -- far more
> > so (isn't it?) than having to (re-)write a ping-like command to call the C
> > code in our base.
>
On Mon, 30 Apr 2007, David Lee wrote:
> [...]
> We already have such code, and already have it duplicated (ouch!) in
> "resources/OCF/IPaddr.in" and "resources/heartbeat/IPaddr.in". And
> "pingd.sh" is in danger of making this triplicate.
> [...]
rsion instead.
The basic underlying idea is to make best efforts for a build-from-source
to work if at all possible. Failure to build should (in theory) only
happen if some sort of "essential" pre-requisite is absent. Absence of a
non-essential should (in theor
ather than
2.0.8, in case there have been any bugfixes in this area since 2.0.8?
http://hg.linux-ha.org/dev/archive/tip.tar.gz
If you wish, send me _off-list_ the entire output of the "configure"
stage, and I could try (alas with no promises) to take a look. (Ideally
from that &quo
On Thu, 3 May 2007, Alan Robertson wrote:
> David Lee wrote:
> > On Mon, 30 Apr 2007, David Lee wrote:
> >
> >> [...]
> >> We already have such code, and already have it duplicated (ouch!) in
> >> "resources/OCF/IPaddr.in" and "resources/h
ommend using heartbeat's "ConfigureMe",
rather than directly using "configure".
Could you check the output of your "ConfigureMe" run (its "configure"
component) to see it discovered about "gnutls" and "quorumd", and how they
may
igureMe" run (its "configure"
> >component) to see it discovered about "gnutls" and "quorumd", and how they
> >may relate to any options passed from "ConfigureMe" to "configure"?
> >Thanks.
It would still be useful,
"/proc", the pid-like
entries seem to be directories. On one of them it is "dr-xr-xr-x", on
another it is "dr-x--x--x".
On the basis of that, would "test -d" be better?
Also, the update has "echo ..." to print configure's output. The official
vant extracts
from configure/build output logs, runtime logfiles, etc., then people here
may be able to offer advice. And if those issues seem to reveal
underlying bugs or documentation holes, we'll try to use that evidence to
fix them.
Hope that helps.
--
: David Lee
annot second guess its location (Linux: /sbin Solaris: /usr/sbin).
How does that seem?
(In the original poster's problem, if "fsck" was present, then configure
should have found it (or there was something flawed with his build process
which needs addressing). If it was absent, th
-2.x series. The official release is 2.0.8,
but I understand that a 2.1.0 is also available and preferred by some
people.
Summary: Use a current OS and a current heartbeat release. Hope that
helps.
--
: David LeeI.T. Service :
: Senior Systems
en in the development tree for some weeks. But I was not easily
able to test the other; that part of bug remains open.
Coincidentally, just this weekend Alan has also visited this area because
of an issue with the fix, which we are now addressing.
Could I suggest that we look at bug 1576, pleas
about building, rather than
running. But it does (if I read it correctly) include the useful
build-time "problem X caused by update Y" functionality.
--
: David LeeI.T. Service :
: Senior Systems Programmer
with the build side of things.)
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComputer Centre :
: UNIX Team Leader Durham University :
: South Road:
:
restart with the revised named.conf configuration (which would need to be
maintained and available across both machines).
When you get it working, it might be worth writing it up as an example for
the heartbeat wiki or other documentation.
Hope that
eam from the function's "stdin" (whose
first-line, first-record becomes the "$1") and to use the proper awk-array
[...] syntax.
Unfortunately when I just tried that, my attempted '[...]' is getting
zapped between "configure.in" and &quo
atch any day now...please...)
I guess that "telecom/recoverymgrd/conf_lex.l" contains a flex-ism;
something that works in GNU/flex but not in manufacturer "lex". If you
could investigate that, and try to provide a fix (that must, of course,
continue to work in "flex&qu
yjob). But I can at least try to reproduce them and
try to narrow down towards the problem area.
Best wishes.
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComputer Centre :
: UNIX Team Leader
, but...)
(There's currently a parallel thread (with Subject typo of "CentoOS")
running about this; I'm also in contact with the CentOS packager.)
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComp
rs for their uid/gid?
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComputer Centre :
: UNIX Team Leader Durham University :
: South Road
en many improvements between those times, including in the
build procedures for non-Linux systems.
If you have problems, whether compile-time or run-time, then report the
details of such problems back here.
Hope that helps.
--
: David LeeI.T. Service :
ce".
Can someone briefly describe what that declaration is wishing to achieve?
Then we can set about understanding it and seeing if it can be made more
reliable.
Hope that helps.
--
: David LeeI.T. Service :
: Senior Systems ProgrammerCompute
aris 10 is an impressive OS, more people should try it.
Actually I tend to stay with S8 and S9. S10 made so many changes to
system boot procedures, and we're locally migrating away from Solaris
towards Linux, that the migration to S10 here seemed a poor choice. (But
I do run a bit of S10 on a c
On Tue, 26 Feb 2008, Dejan Muhamedagic wrote:
> On Tue, Feb 26, 2008 at 01:48:14PM +0000, David Lee wrote:
> > On Mon, 25 Feb 2008 [EMAIL PROTECTED] wrote:
> >
> > > Thanks for the quick responses Dave and Dejan. Dejans suggested fix for
> > > the structure wo
1 to 2). The rebuild
should then work. (I hope!)
To understand what it does, look near line 50 where it is used to pick up
which form of the "reboot(...)" call to use.
Hope that helps.
--
: David LeeI.T. Service :
uot;lib/clplumbing/cl_reboot.c" source file. This has since been properly
fixed in Hg about three weeks ago (Tue Feb 19).
(There have been at least two reports on this list about that, the latest
just earlier today: "heartbeat on solaris 10 (problems)".)
Thanks.
--
:
ad trouble with the 'grandchild' (Solaris) model. Something like that...
I suspect a lurking bug there. If I get time in the next few days I'll see
whether I can refresh my memory of it. (I seem to recall that we
discussed it on the "linux-ha-dev" list.)
Meanwhile, if
find something,
Great! Welcome aboard!
Problem analysis would be welcome. If you can supplement that with
suggested solutions or even bugfixes that would be welcome, but don't
spend disproportionate time on dreaming up actual bugfixes as they can
often require other considerati
On Tue, 15 Apr 2008, David Lee wrote:
> On Tue, 15 Apr 2008, Jan Chaloupecky wrote:
>
> [...]
> > Yes, I read your following email exange in the mailing list archive:
> > http://www.gossamer-threads.com/lists/linuxha/users/40766?search_string=%2Bheartbeat%20%2Bk%20%2BSolar
eartbeat (and pacemaker)
packages themselves built and distributed in CSW/blastwave itself. But
don't hold your breath waiting for that one...)
--
: David LeeI.T. Service :
: Senior Systems Programmer
he list) know your findings (including your
package version, where your package came from etc.).
(The packaging can, in principle, ensure that all dependencies are
fulfilled. But the actual practice (by the packager) of doing this can
sometimes accidentally miss a dependency.)
Hope that helps
On Wed, 28 May 2008, Jan Chaloupecky wrote:
> On Tue, May 27, 2008 at 12:53 PM, David Lee <[EMAIL PROTECTED]> wrote:
> > [...]
> > My suggestion would be to try to use 'csw' for everything, including the
> > 'gcc' compiler.
> >
> &g
eferentially
reporting to "linux-ha-dev" rather than "linux-ha").
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComputer Centre :
: UNIX Team Leader Durham University :
:
if a particular instance failed, the routing would simply find another
instance of that same IP address. (Or, again, something like that.)
--
: David LeeI.T. Service :
: Senior Systems ProgrammerComputer Centre :
: UNIX Tea
ConfigureMe (configure)" output
for stuff to do with the mail program. And check any Makefile to see its
value of "MAILCMD": this ought to be a real pathname, but I suspect might
be empty in your case.
Hope that helps.
--
: Davi
any real meaning) because anycast is purely a
network routing feature which, from heartbeat's persepctive (as an
application sitting on the network) would be "unicast".
Hope that helps.
--
: David LeeI.T. Service :
: Senior Systems Pr
c/
I've no idea whether there is any compatibility overlap between pacemaker
and OHAC. I suspect, sadly, that there might not be (i.e. that the
clustering world has split into two (or more) parts).
Hope that helps a little (even though it is not the simple, bright "yes"
you might ha
Hi, Andrew. Thanks for the reply. Hope you are well.
On Mon, 26 Jan 2009, Andrew Beekhof wrote:
On Mon, Jan 26, 2009 at 18:11, David Lee wrote:
[...]
My guess (but it is
just a guess) is that pacemaker probably no longer builds on non-Linux
machines. Can anyone confirm/refute this
45 matches
Mail list logo