Mailing List Archive: mod_backhand: Wackamole

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8

Index | Next | Previous | View Threaded

minlu_us at yahoo

May 9, 2002, 1:21 PM

Post #1 of 15 (2291 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8

Has anyone successfully built Wackamole on Solaris 8?
I just spent two days trying to get it to compile in
vain. There is no documentation. Where can I get the
binary of Wackamole if I just can't build it?

Thanks.

Min

__________________________________________________
Do You Yahoo!?
Yahoo! Shopping - Mother's Day is May 12th!
http://shopping.yahoo.com

alee at 2win

May 9, 2002, 2:34 PM

Post #2 of 15 (2231 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

I was able to get it compiled but still can't get it to work. The address
didn't get taken over by the backup machine as advertised.
I am also looking to hear from someone who had successfully got it to work
on Solaris 8. I would really like to use wackamole with mod_backhand for
load balancing and fault-tolerance. Anyone?

Alex

On Thu, 9 May 2002, Min Lu wrote:

>
> Has anyone successfully built Wackamole on Solaris 8?
> I just spent two days trying to get it to compile in
> vain. There is no documentation. Where can I get the
> binary of Wackamole if I just can't build it?
>
> Thanks.
>
> Min
>
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Shopping - Mother's Day is May 12th!
> http://shopping.yahoo.com
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>

munjal at commedia

May 9, 2002, 4:46 PM

Post #3 of 15 (2232 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Hi!

Thank you for your interest in Wackamole. Unfortunately, we do not
have access to a Solaris 8 box at the CNDS lab (http://www.cnds.jhu.edu)
at Hopkins, so we cannot currently test it for this architecture.
Most of the Wackamole code is very protable and should work on
any operating system, similar to the Spread toolkit. However, the
part that actually grabs the IPs is VERY system dependent, potentially
to the level of certain kernels.

We could look into the problem if someone can give us access to
a Solaris 8 environment.

Aashima

On Thu, 9 May 2002, Alex Lee wrote:

>
> I was able to get it compiled but still can't get it to work. The address
> didn't get taken over by the backup machine as advertised.
> I am also looking to hear from someone who had successfully got it to work
> on Solaris 8. I would really like to use wackamole with mod_backhand for
> load balancing and fault-tolerance. Anyone?
>
> Alex
>
> On Thu, 9 May 2002, Min Lu wrote:
>
> >
> > Has anyone successfully built Wackamole on Solaris 8?
> > I just spent two days trying to get it to compile in
> > vain. There is no documentation. Where can I get the
> > binary of Wackamole if I just can't build it?
> >
> > Thanks.
> >
> > Min
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Yahoo! Shopping - Mother's Day is May 12th!
> > http://shopping.yahoo.com
> >
> > _______________________________________________
> > wackamole-users mailing list
> > wackamole-users [at] lists
> > http://lists.backhand.org/mailman/listinfo/wackamole-users
> >
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>

yairamir at cnds

May 9, 2002, 5:48 PM

Post #4 of 15 (2231 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Hey,

If someone is even able to spare a solaris 8 box, I am sure
we will be happy to find a power outlet for it in the computer room
of the lab :)

Cheers,

:) Yair.

Aashima Munjal wrote:

> Hi!
>
> Thank you for your interest in Wackamole. Unfortunately, we do not
> have access to a Solaris 8 box at the CNDS lab (http://www.cnds.jhu.edu)
> at Hopkins, so we cannot currently test it for this architecture.
> Most of the Wackamole code is very protable and should work on
> any operating system, similar to the Spread toolkit. However, the
> part that actually grabs the IPs is VERY system dependent, potentially
> to the level of certain kernels.
>
> We could look into the problem if someone can give us access to
> a Solaris 8 environment.
>
> Aashima
>
>
>
> On Thu, 9 May 2002, Alex Lee wrote:
>
>
>>I was able to get it compiled but still can't get it to work. The address
>>didn't get taken over by the backup machine as advertised.
>>I am also looking to hear from someone who had successfully got it to work
>>on Solaris 8. I would really like to use wackamole with mod_backhand for
>>load balancing and fault-tolerance. Anyone?
>>
>>Alex
>>
>>On Thu, 9 May 2002, Min Lu wrote:
>>
>>
>>>Has anyone successfully built Wackamole on Solaris 8?
>>>I just spent two days trying to get it to compile in
>>>vain. There is no documentation. Where can I get the
>>>binary of Wackamole if I just can't build it?
>>>
>>>Thanks.
>>>
>>>Min
>>>
>>>__________________________________________________
>>>Do You Yahoo!?
>>>Yahoo! Shopping - Mother's Day is May 12th!
>>>http://shopping.yahoo.com
>>>
>>>_______________________________________________
>>>wackamole-users mailing list
>>>wackamole-users [at] lists
>>>http://lists.backhand.org/mailman/listinfo/wackamole-users
>>>
>>>
>>
>>_______________________________________________
>>wackamole-users mailing list
>>wackamole-users [at] lists
>>http://lists.backhand.org/mailman/listinfo/wackamole-users
>>
>>
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>
>

minlu_us at yahoo

May 9, 2002, 7:37 PM

Post #5 of 15 (2229 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Thanks for all quick quick reponse. I really
appreciated. As we all know, almost all current
commercial HA implementation have propriatary
components in them. This wackamole thing looks
have pretty good abstraction to me.

Alex,

Can you share some idea with me? Do you have
a makefile? Yair, I will look and see if I
can find a box. You guys are great. It reminds me
of my graduate school days ten years ago.

Thanks.

Min

--- Yair Amir <yairamir [at] cnds> wrote:
> Hey,
>
> If someone is even able to spare a solaris 8 box, I
> am sure
> we will be happy to find a power outlet for it in
> the computer room
> of the lab :)
>
> Cheers,
>
> :) Yair.
>
>
> Aashima Munjal wrote:
>
> > Hi!
> >
> > Thank you for your interest in Wackamole.
> Unfortunately, we do not
> > have access to a Solaris 8 box at the CNDS lab
> (http://www.cnds.jhu.edu)
> > at Hopkins, so we cannot currently test it for
> this architecture.
> > Most of the Wackamole code is very protable and
> should work on
> > any operating system, similar to the Spread
> toolkit. However, the
> > part that actually grabs the IPs is VERY system
> dependent, potentially
> > to the level of certain kernels.
> >
> > We could look into the problem if someone can give
> us access to
> > a Solaris 8 environment.
> >
> > Aashima
> >
> >
> >
> > On Thu, 9 May 2002, Alex Lee wrote:
> >
> >
> >>I was able to get it compiled but still can't get
> it to work. The address
> advertised.
> successfully got it to work
> with mod_backhand for
> Solaris 8?
> in
> the
>
>>>__________________________________________________
>
>>>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>>http://lists.backhand.org/mailman/listinfo/wackamole-users
> >>
> >>
> >
> >
> > _______________________________________________
> > wackamole-users mailing list
> > wackamole-users [at] lists
> >
>
http://lists.backhand.org/mailman/listinfo/wackamole-users
> >
> >
> >
>
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
>
http://lists.backhand.org/mailman/listinfo/wackamole-users

__________________________________________________
Do You Yahoo!?
Yahoo! Shopping - Mother's Day is May 12th!
http://shopping.yahoo.com

aroutley at team

May 9, 2002, 7:53 PM

Post #6 of 15 (2217 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Min,

I haven't got much time right at the moment, but I should be able
to compile it and provide you with install notes in a few hours.

Where abouts is the compile breaking for you at the moment?

Later
Andrew

> -----Original Message-----
> From: Min Lu [mailto:minlu_us [at] yahoo]
> Sent: Friday, 10 May 2002 12:38
> To: wackamole-users [at] lists
> Subject: Re: [Wackamole-users] Having a terrible time build
> wackamole_CVS_200203211357 on Solaris 8
>=20
>=20
>=20
> Thanks for all quick quick reponse. I really
> appreciated. As we all know, almost all current
> commercial HA implementation have propriatary
> components in them. This wackamole thing looks
> have pretty good abstraction to me.
>=20
> Alex,
>=20
> Can you share some idea with me? Do you have
> a makefile? Yair, I will look and see if I
> can find a box. You guys are great. It reminds me
> of my graduate school days ten years ago.
>=20
> Thanks.
>=20
> Min
>=20
>=20
>=20
>=20
>=20
> --- Yair Amir <yairamir [at] cnds> wrote:
> > Hey,
> >=20
> > If someone is even able to spare a solaris 8 box, I
> > am sure
> > we will be happy to find a power outlet for it in
> > the computer room
> > of the lab :)
> >=20
> > Cheers,
> >=20
> > :) Yair.
> >=20
> >=20
> > Aashima Munjal wrote:
> >=20
> > > Hi!
> > >=20
> > > Thank you for your interest in Wackamole.
> > Unfortunately, we do not
> > > have access to a Solaris 8 box at the CNDS lab
> > (http://www.cnds.jhu.edu)
> > > at Hopkins, so we cannot currently test it for
> > this architecture.
> > > Most of the Wackamole code is very protable and
> > should work on
> > > any operating system, similar to the Spread
> > toolkit. However, the
> > > part that actually grabs the IPs is VERY system
> > dependent, potentially
> > > to the level of certain kernels.
> > >=20
> > > We could look into the problem if someone can give
> > us access to
> > > a Solaris 8 environment.
> > >=20
> > > Aashima
> > >=20
> > >=20
> > >=20
> > > On Thu, 9 May 2002, Alex Lee wrote:
> > >=20
> > >=20
> > >>I was able to get it compiled but still can't get
> > it to work. The address
> > advertised.
> > successfully got it to work
> > with mod_backhand for
> > Solaris 8?
> > in
> > the
> >
> >>>__________________________________________________
> >
> >>>http://lists.backhand.org/mailman/listinfo/wackamole-users
> >
> >>http://lists.backhand.org/mailman/listinfo/wackamole-users
> > >>
> > >>
> > >=20
> > >=20
> > > _______________________________________________
> > > wackamole-users mailing list
> > > wackamole-users [at] lists
> > >
> >
> http://lists.backhand.org/mailman/listinfo/wackamole-users
> > _______________________________________________
> > wackamole-users mailing list
> > wackamole-users [at] lists
> >
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>=20
>=20
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Shopping - Mother's Day is May 12th!
> http://shopping.yahoo.com
>=20
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>=20

minlu_us at yahoo

May 9, 2002, 8:31 PM

Post #7 of 15 (2216 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Since I can see Wackamole in action right now,
can you tell me if my understanding is correct
please?

Most LB-type of clusters requires a up-front
function either in the form of hardware(L4 switches)
or software(Resonate, etc) that spreads the ips(
translate virtual IP to physical IP of the nodes.

This Wackamole arbitrates the leader in the membership
amongst the nodes themselves. I assume there is no
shared quorum between them, will IP takeover always
be successful? Particularly when all nodes are down
and
coming back at the same time or "split-brain"?

Any advice is greatly appreciated.

Min

--- Yair Amir <yairamir [at] cnds> wrote:
> Hey,
>
> If someone is even able to spare a solaris 8 box, I
> am sure
> we will be happy to find a power outlet for it in
> the computer room
> of the lab :)
>
> Cheers,
>
> :) Yair.
>
>
> Aashima Munjal wrote:
>
> > Hi!
> >
> > Thank you for your interest in Wackamole.
> Unfortunately, we do not
> > have access to a Solaris 8 box at the CNDS lab
> (http://www.cnds.jhu.edu)
> > at Hopkins, so we cannot currently test it for
> this architecture.
> > Most of the Wackamole code is very protable and
> should work on
> > any operating system, similar to the Spread
> toolkit. However, the
> > part that actually grabs the IPs is VERY system
> dependent, potentially
> > to the level of certain kernels.
> >
> > We could look into the problem if someone can give
> us access to
> > a Solaris 8 environment.
> >
> > Aashima
> >
> >
> >
> > On Thu, 9 May 2002, Alex Lee wrote:
> >
> >
> >>I was able to get it compiled but still can't get
> it to work. The address
> advertised.
> successfully got it to work
> with mod_backhand for
> Solaris 8?
> in
> the
>
>>>__________________________________________________
>
>>>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>>http://lists.backhand.org/mailman/listinfo/wackamole-users
> >>
> >>
> >
> >
> > _______________________________________________
> > wackamole-users mailing list
> > wackamole-users [at] lists
> >
>
http://lists.backhand.org/mailman/listinfo/wackamole-users
> >
> >
> >
>
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
>
http://lists.backhand.org/mailman/listinfo/wackamole-users

__________________________________________________
Do You Yahoo!?
Yahoo! Shopping - Mother's Day is May 12th!
http://shopping.yahoo.com

yairamir at cnds

May 9, 2002, 9:25 PM

Post #8 of 15 (2229 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Hi,

Min Lu wrote:

> Since I can see Wackamole in action right now,
> can you tell me if my understanding is correct
> please?
>
> Most LB-type of clusters requires a up-front
> function either in the form of hardware(L4 switches)
> or software(Resonate, etc) that spreads the ips(
> translate virtual IP to physical IP of the nodes.
>
> This Wackamole arbitrates the leader in the membership
> amongst the nodes themselves. I assume there is no
> shared quorum between them, will IP takeover always
> be successful? Particularly when all nodes are down
> and
> coming back at the same time or "split-brain"?

Assuming the Spread daemons and the Wackamole processes work,
there should be no problem. Wackamole relies on the Spread membership
service to know who is alive and connected. This will be in agreement
among all those that are connected. Now, this is a bit delicate
and to fully understand it you will need to understand the
membership semantics Spread provides.

One of the reasons for me to write Wackamole was to demonstrate
how a correct application for the Spread toolkit should be written.
(Other people, of course, had much better reasons - such as writting
cool hacking code that can create havoc in the lab :)

The consistent state engine in Wackamole demonstrates how to
write consistent state agreement in partitionable environments
(environments that can partition to several components or merge,
where ANY events, possibly cascading, of membership changes is
possible). This takes care of any "split-brain" problem, as all
the IP addresses will be covered exactly once, in each of the
partitioned components that contain at least one Wackamole running.

Enjoy,

:) Yair.

aroutley at team

May 9, 2002, 10:53 PM

Post #9 of 15 (2239 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

>=20
> Has anyone successfully built Wackamole on Solaris 8?
> I just spent two days trying to get it to compile in
> vain. There is no documentation. Where can I get the
> binary of Wackamole if I just can't build it?
>=20
ok, I've had time to have a look at the source now, and there
are two problems making it impossible to create the final
binary (all the source comiples relatively cleanly):

1. There is no code for sample_arp_cache for Solaris 8.
It neither compiles under the HAVE_PROC_NET_ARP nor the
CTL_NET define requirements. This function is called from
wackamole.c

2. The same kind of problem for serv_accept, Solaris defines
I_PUSH and under the apue.c defines this function doesn't
get created and control.c requires it.

This kind of programming is well beyond my admin knowledge so
although I could force it to compile relatively cleanly I can't
guarantee the resulting behaviour :-)

Changes required to get the source files compiled:
1. Require GNU Bison and flex
2. After running configure I edited the Makefile and removed the
-I/usr/include from CPPFLAGS (causes alarm.c not to compile)
3. Edited control.c and added #include <sys/filio.h> so that FIOBIN
was defined, otherwise control.c does not compile.

Finally failed with:
gcc -o wackamole wackamole.o alarm.o lex.yy.o y.tab.o arpcache.o
spoofmanager.o abt.o control.o apue.o configuration.o ife-dlpi.o
-L/usr/local/lib -R/usr/local/lib -ltsp -ldl -lsocket -lnsl -lposix4
-lthread -lpthread
Undefined first referenced
symbol in file
serv_accept control.o
sample_arp_cache wackamole.o

Later
Andrew

jesus at omniti

May 10, 2002, 4:05 PM

Post #10 of 15 (2227 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Alex Lee wrote:

>I was able to get it compiled but still can't get it to work. The address
>didn't get taken over by the backup machine as advertised.
>I am also looking to hear from someone who had successfully got it to work
>on Solaris 8. I would really like to use wackamole with mod_backhand for
>load balancing and fault-tolerance. Anyone?
>

All of the Solaris 8 support for wackamole was added in the CVS version
and tested on Solaris 8 Intel. The Intel branch of Solaris has an
identical userland interface/API for all system stuff. So, it should
work fine on Solaris 8 Sparc.

I don't believe that the release version has any hope of running on
Solaris 8... The OS specific stuff simply isn't finished there.

--
Theo Schlossnagle
Principal Consultant
OmniTI Computer Consulting, Inc. -- http://www.omniti.com/
Phone: +1 301 776 6376 Fax: +1 410 880 4879
1024D/82844984/95FD 30F1 489E 4613 F22E 491A 7E88 364C 8284 4984
2047R/33131B65/71 F7 95 64 49 76 5D BA 3D 90 B9 9F BE 27 24 E7

jesus at omniti

May 10, 2002, 4:19 PM

Post #11 of 15 (2238 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Andrew Routley wrote:

>>ok, I've had time to have a look at the source now, and there
>>are two problems making it impossible to create the final
>>binary (all the source comiples relatively cleanly):
>>
>>1. There is no code for sample_arp_cache for Solaris 8.
>> It neither compiles under the HAVE_PROC_NET_ARP nor the
>> CTL_NET define requirements. This function is called from
>> wackamole.c
>>
This means that you will need to explicitly specify which addresses to
arp spoof to. This is the way wackamole release version is anyway, but
here you can specify which interface and entire subnets. That code for
Solaris is unfinished. I will try to get to it soon.

>>2. The same kind of problem for serv_accept, Solaris defines
>> I_PUSH and under the apue.c defines this function doesn't
>> get created and control.c requires it.
>>
I have code for this... I just need an hour or so on a Solaris box to
clean that up.

>>This kind of programming is well beyond my admin knowledge so
>>although I could force it to compile relatively cleanly I can't
>>guarantee the resulting behaviour :-)
>>
Mine too apparently :-D

>>Changes required to get the source files compiled:
>>1. Require GNU Bison and flex
>>
Yup. The next release will not, but the CVS version will. flex and
bison will be run before the release source is packaged... So you won't
need them.

>>2. After running configure I edited the Makefile and removed the
>> -I/usr/include from CPPFLAGS (causes alarm.c not to compile)
>>
Can you mail be the compile error please?

>>3. Edited control.c and added #include <sys/filio.h> so that FIOBIN
>> was defined, otherwise control.c does not compile.
>>
Thanks.. I just added that to the autoconf stuff.

>>
>>Finally failed with:
>>gcc -o wackamole wackamole.o alarm.o lex.yy.o y.tab.o arpcache.o
>>spoofmanager.o abt.o control.o apue.o configuration.o ife-dlpi.o
>>-L/usr/local/lib -R/usr/local/lib -ltsp -ldl -lsocket -lnsl -lposix4
>>-lthread -lpthread
>>Undefined first referenced
>> symbol in file
>>serv_accept control.o
>>sample_arp_cache wackamole.o
>>
The first is a quick fix on my end. I need to put that code into CVS.
The second is a quick fix, if you do'nt want a real arp cache. Just
stub the function out :-) [.I just committed a tub to CVS -- I'll have
it do the right thing on Soalris as soon as I figure out how]

Sorry for all the trouble (you too Min). I am running this stuff on
FreeBSD actively, so that always gets the first bit of attention :-)

--
Theo Schlossnagle
Principal Consultant
OmniTI Computer Consulting, Inc. -- http://www.omniti.com/
Phone: +1 301 776 6376 Fax: +1 410 880 4879
1024D/82844984/95FD 30F1 489E 4613 F22E 491A 7E88 364C 8284 4984
2047R/33131B65/71 F7 95 64 49 76 5D BA 3D 90 B9 9F BE 27 24 E7

minlu_us at yahoo

May 22, 2002, 9:27 AM

Post #12 of 15 (2231 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Yair,

Thanks for your response. Now it's time you
toot your honk a little bit. Where can I
find some of the current users of Spread
and Wackamole? I thought I came across
something earlier, but don't remember where.

Thanks,

Min

--- Yair Amir <yairamir [at] cnds> wrote:
> Hi,
>
> Min Lu wrote:
>
> > Since I can see Wackamole in action right now,
> > can you tell me if my understanding is correct
> > please?
> >
> > Most LB-type of clusters requires a up-front
> > function either in the form of hardware(L4
> switches)
> > or software(Resonate, etc) that spreads the ips(
> > translate virtual IP to physical IP of the nodes.
> >
> > This Wackamole arbitrates the leader in the
> membership
> > amongst the nodes themselves. I assume there is no
> > shared quorum between them, will IP takeover
> always
> > be successful? Particularly when all nodes are
> down
> > and
> > coming back at the same time or "split-brain"?
>
>
> Assuming the Spread daemons and the Wackamole
> processes work,
> there should be no problem. Wackamole relies on the
> Spread membership
> service to know who is alive and connected. This
> will be in agreement
> among all those that are connected. Now, this is a
> bit delicate
> and to fully understand it you will need to
> understand the
> membership semantics Spread provides.
>
> One of the reasons for me to write Wackamole was to
> demonstrate
> how a correct application for the Spread toolkit
> should be written.
> (Other people, of course, had much better reasons -
> such as writting
> cool hacking code that can create havoc in the lab
> :)
>
> The consistent state engine in Wackamole
> demonstrates how to
> write consistent state agreement in partitionable
> environments
> (environments that can partition to several
> components or merge,
> where ANY events, possibly cascading, of membership
> changes is
> possible). This takes care of any "split-brain"
> problem, as all
> the IP addresses will be covered exactly once, in
> each of the
> partitioned components that contain at least one
> Wackamole running.
>
> Enjoy,
>
> :) Yair.
>
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
>
http://lists.backhand.org/mailman/listinfo/wackamole-users

__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com

jroysdon at hotmail

May 28, 2002, 2:30 PM

Post #13 of 15 (2228 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

I've emailed this to Yair and Aashima, but I figured I'd post it on the list
in case this makes it to them any faster, plus just let the rest of you know
my intentions. My original email to Dr. Amir:

>Dr. Amir,
>
> First, I want to make sure I'm sending this to the right contact.
> You
>were listed as the first contact on the Wackamole page. However, if you
> know
>of a more appropriate contact or resource list for me to consult, I'm
> all
>ears.
>
>
>
>I've been working on using Spread and Wackamole to replicate some
>
>front-end services. I've managed to get both apps installed and working
>
>successfully under RedHat Linux, but I'm running into problems with
>
>SunOS 5.8.
>
>
>
> I have Spread installed on the SunOS server and communicating with
> the
>two existing Linux servers. However, when I start Wackamole, it
>
>communicates via Spread to the other servers the it's going to acquire it's
>
>VIP and whichever server was running a VIP disables its interface, but the
>
>SunOS server never binds a virtual address. I verify this with ifconfig -a
>
>and only the real IP is bound (unlike in Linux where I'd see the real and
>
>VIPs bound).
>
>
>
> Any suggestions or hints as to what to look at? When I run
>
>Wackamole in non-daemonized mode I see the following:
>
>
>
>alias up failed
>
>Aquire address 192.168.64.25
>
>
>
> However, if I manaully bind the VIP, Wackamole is able to unbind it:
>
>
> Manual
>bind:
> #ifconfig
>dmfe0:1 plumb 192.168.64.24 netmask 255.255.255.128 broadcast
>
>192.168.64.127 up
> #
>ifconfig -a
> lo0:
>flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
>
>inet 127.0.0.1 netmask ff000000
> dmfe0:
>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>
>inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
>
>ether 0:3:ba:11:f0:b1
> dmfe0:1:
>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>
>inet 192.168.64.24 netmask ffffff80 broadcast 192.168.64.127
>
>
> Wackamole
>sees another Linux server load with VIP priority:
>
>Priority_claim called
>
>if_up: 192.168.64.25 0.0.0.0 0.0.0.0
>
>Release address 192.168.64.25
>
>Interface truely unbound:
> #
>ifconfig -a
> lo0:
>flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
>
>inet 127.0.0.1 netmask ff000000
> dmfe0:
>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>
>inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
>
>ether 0:3:ba:11:f0:b1
>
>
> Thanks in
>advance for any help, hints, or direction you're able to give,

My followup email after reading Aashima & Yair's posts:

>Ah, poking around backhand.org I found the wackamole user list! I see that
>
>others have reported this same problem and you're in need of a Solaris
>Sparc
> environment to test with.
>
>
>
>I'd gladly let you use two of my Sun Netra X1 in order to get the ball
>
>rolling again for SunOS/Solaris. These are headless servers with out of
>
>band power management (Sun calls it "LOM") and direct console access via a
>
>Cisco terminal server, as well as ssh access when you're not hosing the
>
>network interface.
>
>
> Just
>let me know who to contact to get them access. I am running a very
> secure
>enviroment here, but I can VLAN off two of the servers and allow
> access
>through our firewall to our Cisco TS to a few folks.
>
>
> For
>security, I would prefer if either you can access our network via a VPN
> or SSH
>and then telnet locally to the Cisco TS for LOM power management and
> console
>access.

Jason Roysdon, CCDP/CCNP:Security, MCSE, CNA, Network+, A+
Public email : jroysdon [at] hotmail
Homepage : http://jason.roysdon.net/

PS: Don't use a subject like that looks like spam, as your message will end
up in my junk filter and never be read.

Yair Amir Thu, 09 May 2002 20:48:29 -0400:

Hey,

If someone is even able to spare a solaris 8 box, I am sure
we will be happy to find a power outlet for it in the computer room
of the lab :)

Cheers,

:) Yair.

Aashima Munjal wrote:

>Hi!
> > Thank you for your interest in Wackamole. Unfortunately, we do not
>have access to a Solaris 8 box at the CNDS lab (http://www.cnds.jhu.edu)
>at Hopkins, so we cannot currently test it for this architecture.
>Most of the Wackamole code is very protable and should work on
>any operating system, similar to the Spread toolkit. However, the
>part that actually grabs the IPs is VERY system dependent, potentially
>to the level of certain kernels.
> > We could look into the problem if someone can give us access to
>a Solaris 8 environment.
> > Aashima
> > > > On Thu, 9 May 2002, Alex Lee wrote:
> > >>I was able to get it compiled but still can't get it to work. The
>address
>>didn't get taken over by the backup machine as advertised.
>>I am also looking to hear from someone who had successfully got it to work
>>on Solaris 8. I would really like to use wackamole with mod_backhand for
>>load balancing and fault-tolerance. Anyone?
>>
>>Alex
>>
>>On Thu, 9 May 2002, Min Lu wrote:
>>
>>
>>>Has anyone successfully built Wackamole on Solaris 8?
>>>I just spent two days trying to get it to compile in
>>>vain. There is no documentation. Where can I get the
>>>binary of Wackamole if I just can't build it?
>>>
>>>Thanks.
>>>
>>>Min
>>>

_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com

alee at 2win

May 28, 2002, 2:56 PM

Post #14 of 15 (2228 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

Jason,

I got pretty much similar problem in my Solaris 8 installation. I am
glad to see that you can offer two Solaris boxes for troubleshooting. Has
the offer been taken by Wackamole developers? I think quite a few of us
are anxiously looking forward to be able to run spread/wackamole/backhand
on Solaris.

Thanks!

Alex

On Tue, 28 May 2002, Jason Roysdon wrote:

> I've emailed this to Yair and Aashima, but I figured I'd post it on the list
> in case this makes it to them any faster, plus just let the rest of you know
> my intentions. My original email to Dr. Amir:
>
> >Dr. Amir,
> >
> > First, I want to make sure I'm sending this to the right contact.
> > You
> >were listed as the first contact on the Wackamole page. However, if you
> > know
> >of a more appropriate contact or resource list for me to consult, I'm
> > all
> >ears.
> >
> >
> >
> >I've been working on using Spread and Wackamole to replicate some
> >
> >front-end services. I've managed to get both apps installed and working
> >
> >successfully under RedHat Linux, but I'm running into problems with
> >
> >SunOS 5.8.
> >
> >
> >
> > I have Spread installed on the SunOS server and communicating with
> > the
> >two existing Linux servers. However, when I start Wackamole, it
> >
> >communicates via Spread to the other servers the it's going to acquire it's
> >
> >VIP and whichever server was running a VIP disables its interface, but the
> >
> >SunOS server never binds a virtual address. I verify this with ifconfig -a
> >
> >and only the real IP is bound (unlike in Linux where I'd see the real and
> >
> >VIPs bound).
> >
> >
> >
> > Any suggestions or hints as to what to look at? When I run
> >
> >Wackamole in non-daemonized mode I see the following:
> >
> >
> >
> >alias up failed
> >
> >Aquire address 192.168.64.25
> >
> >
> >
> > However, if I manaully bind the VIP, Wackamole is able to unbind it:
> >
> >
> > Manual
> >bind:
> > #ifconfig
> >dmfe0:1 plumb 192.168.64.24 netmask 255.255.255.128 broadcast
> >
> >192.168.64.127 up
> > #
> >ifconfig -a
> > lo0:
> >flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
> >
> >inet 127.0.0.1 netmask ff000000
> > dmfe0:
> >flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
> >
> >inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
> >
> >ether 0:3:ba:11:f0:b1
> > dmfe0:1:
> >flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
> >
> >inet 192.168.64.24 netmask ffffff80 broadcast 192.168.64.127
> >
> >
> > Wackamole
> >sees another Linux server load with VIP priority:
> >
> >Priority_claim called
> >
> >if_up: 192.168.64.25 0.0.0.0 0.0.0.0
> >
> >Release address 192.168.64.25
> >
> >Interface truely unbound:
> > #
> >ifconfig -a
> > lo0:
> >flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
> >
> >inet 127.0.0.1 netmask ff000000
> > dmfe0:
> >flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
> >
> >inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
> >
> >ether 0:3:ba:11:f0:b1
> >
> >
> > Thanks in
> >advance for any help, hints, or direction you're able to give,
>
> My followup email after reading Aashima & Yair's posts:
>
> >Ah, poking around backhand.org I found the wackamole user list! I see that
> >
> >others have reported this same problem and you're in need of a Solaris
> >Sparc
> > environment to test with.
> >
> >
> >
> >I'd gladly let you use two of my Sun Netra X1 in order to get the ball
> >
> >rolling again for SunOS/Solaris. These are headless servers with out of
> >
> >band power management (Sun calls it "LOM") and direct console access via a
> >
> >Cisco terminal server, as well as ssh access when you're not hosing the
> >
> >network interface.
> >
> >
> > Just
> >let me know who to contact to get them access. I am running a very
> > secure
> >enviroment here, but I can VLAN off two of the servers and allow
> > access
> >through our firewall to our Cisco TS to a few folks.
> >
> >
> > For
> >security, I would prefer if either you can access our network via a VPN
> > or SSH
> >and then telnet locally to the Cisco TS for LOM power management and
> > console
> >access.
>
>
> Jason Roysdon, CCDP/CCNP:Security, MCSE, CNA, Network+, A+
> Public email : jroysdon [at] hotmail
> Homepage : http://jason.roysdon.net/
>
> PS: Don't use a subject like that looks like spam, as your message will end
> up in my junk filter and never be read.
>
>
> Yair Amir Thu, 09 May 2002 20:48:29 -0400:
>
> Hey,
>
> If someone is even able to spare a solaris 8 box, I am sure
> we will be happy to find a power outlet for it in the computer room
> of the lab :)
>
> Cheers,
>
> :) Yair.
>
>
> Aashima Munjal wrote:
>
> >Hi!
> > > Thank you for your interest in Wackamole. Unfortunately, we do not
> >have access to a Solaris 8 box at the CNDS lab (http://www.cnds.jhu.edu)
> >at Hopkins, so we cannot currently test it for this architecture.
> >Most of the Wackamole code is very protable and should work on
> >any operating system, similar to the Spread toolkit. However, the
> >part that actually grabs the IPs is VERY system dependent, potentially
> >to the level of certain kernels.
> > > We could look into the problem if someone can give us access to
> >a Solaris 8 environment.
> > > Aashima
> > > > > On Thu, 9 May 2002, Alex Lee wrote:
> > > >>I was able to get it compiled but still can't get it to work. The
> >address
> >>didn't get taken over by the backup machine as advertised.
> >>I am also looking to hear from someone who had successfully got it to work
> >>on Solaris 8. I would really like to use wackamole with mod_backhand for
> >>load balancing and fault-tolerance. Anyone?
> >>
> >>Alex
> >>
> >>On Thu, 9 May 2002, Min Lu wrote:
> >>
> >>
> >>>Has anyone successfully built Wackamole on Solaris 8?
> >>>I just spent two days trying to get it to compile in
> >>>vain. There is no documentation. Where can I get the
> >>>binary of Wackamole if I just can't build it?
> >>>
> >>>Thanks.
> >>>
> >>>Min
> >>>
>
> _________________________________________________________________
> Chat with friends online, try MSN Messenger: http://messenger.msn.com
>
>
> _______________________________________________
> wackamole-users mailing list
> wackamole-users [at] lists
> http://lists.backhand.org/mailman/listinfo/wackamole-users
>

yairamir at cnds

May 28, 2002, 4:23 PM

Post #15 of 15 (2330 views)
Permalink

Having a terrible time build wackamole_CVS_200203211357 on Solaris 8 [In reply to]

As we indicated a few weeks ago:
If someone provides access to the environment we can have a look.

I am sure Aashima can call you sometime to set access. Bare in mind
these programs really play with the IP addresses, so make sure your
network will not get crazy because of that (our does sometimes when we
debug).

:) Yair.

Jason> I've emailed this to Yair and Aashima, but I figured I'd post it on the list
Jason> in case this makes it to them any faster, plus just let the rest of you know
Jason> my intentions. My original email to Dr. Amir:

>>Dr. Amir,
>>
>> First, I want to make sure I'm sending this to the right contact.
>> You
>>were listed as the first contact on the Wackamole page. However, if you
>> know
>>of a more appropriate contact or resource list for me to consult, I'm
>> all
>>ears.
>>
>>
>>
>>I've been working on using Spread and Wackamole to replicate some
>>
>>front-end services. I've managed to get both apps installed and working
>>
>>successfully under RedHat Linux, but I'm running into problems with
>>
>>SunOS 5.8.
>>
>>
>>
>> I have Spread installed on the SunOS server and communicating with
>> the
>>two existing Linux servers. However, when I start Wackamole, it
>>
>>communicates via Spread to the other servers the it's going to acquire it's
>>
>>VIP and whichever server was running a VIP disables its interface, but the
>>
>>SunOS server never binds a virtual address. I verify this with ifconfig -a
>>
>>and only the real IP is bound (unlike in Linux where I'd see the real and
>>
>>VIPs bound).
>>
>>
>>
>> Any suggestions or hints as to what to look at? When I run
>>
>>Wackamole in non-daemonized mode I see the following:
>>
>>
>>
>>alias up failed
>>
>>Aquire address 192.168.64.25
>>
>>
>>
>> However, if I manaully bind the VIP, Wackamole is able to unbind it:
>>
>>
>> Manual
>>bind:
>> #ifconfig
>>dmfe0:1 plumb 192.168.64.24 netmask 255.255.255.128 broadcast
>>
>>192.168.64.127 up
>> #
>>ifconfig -a
>> lo0:
>>flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
>>
>>inet 127.0.0.1 netmask ff000000
>> dmfe0:
>>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>>
>>inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
>>
>>ether 0:3:ba:11:f0:b1
>> dmfe0:1:
>>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>>
>>inet 192.168.64.24 netmask ffffff80 broadcast 192.168.64.127
>>
>>
>> Wackamole
>>sees another Linux server load with VIP priority:
>>
>>Priority_claim called
>>
>>if_up: 192.168.64.25 0.0.0.0 0.0.0.0
>>
>>Release address 192.168.64.25
>>
>>Interface truely unbound:
>> #
>>ifconfig -a
>> lo0:
>>flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
>>
>>inet 127.0.0.1 netmask ff000000
>> dmfe0:
>>flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
>>
>>inet 192.168.64.4 netmask ffffff80 broadcast 192.168.64.127
>>
>>ether 0:3:ba:11:f0:b1
>>
>>
>> Thanks in
>>advance for any help, hints, or direction you're able to give,

Jason> My followup email after reading Aashima & Yair's posts:

>>Ah, poking around backhand.org I found the wackamole user list! I see that
>>
>>others have reported this same problem and you're in need of a Solaris
>>Sparc
>> environment to test with.
>>
>>
>>
>>I'd gladly let you use two of my Sun Netra X1 in order to get the ball
>>
>>rolling again for SunOS/Solaris. These are headless servers with out of
>>
>>band power management (Sun calls it "LOM") and direct console access via a
>>
>>Cisco terminal server, as well as ssh access when you're not hosing the
>>
>>network interface.
>>
>>
>> Just
>>let me know who to contact to get them access. I am running a very
>> secure
>>enviroment here, but I can VLAN off two of the servers and allow
>> access
>>through our firewall to our Cisco TS to a few folks.
>>
>>
>> For
>>security, I would prefer if either you can access our network via a VPN
>> or SSH
>>and then telnet locally to the Cisco TS for LOM power management and
>> console
>>access.

Jason> Jason Roysdon, CCDP/CCNP:Security, MCSE, CNA, Network+, A+
Jason> Public email : jroysdon [at] hotmail
Jason> Homepage : http://jason.roysdon.net/

Jason> PS: Don't use a subject like that looks like spam, as your message will end
Jason> up in my junk filter and never be read.

Jason> Yair Amir Thu, 09 May 2002 20:48:29 -0400:

Jason> Hey,

Jason> If someone is even able to spare a solaris 8 box, I am sure
Jason> we will be happy to find a power outlet for it in the computer room
Jason> of the lab :)

Jason> Cheers,

Jason> :) Yair.

Jason> Aashima Munjal wrote:

>>Hi!
>> > Thank you for your interest in Wackamole. Unfortunately, we do not
>>have access to a Solaris 8 box at the CNDS lab (http://www.cnds.jhu.edu)
>>at Hopkins, so we cannot currently test it for this architecture.
>>Most of the Wackamole code is very protable and should work on
>>any operating system, similar to the Spread toolkit. However, the
>>part that actually grabs the IPs is VERY system dependent, potentially
>>to the level of certain kernels.
>> > We could look into the problem if someone can give us access to
>>a Solaris 8 environment.
>> > Aashima
>> > > > On Thu, 9 May 2002, Alex Lee wrote:
>> > >>I was able to get it compiled but still can't get it to work. The
>>address
>>>didn't get taken over by the backup machine as advertised.
>>>I am also looking to hear from someone who had successfully got it to work
>>>on Solaris 8. I would really like to use wackamole with mod_backhand for
>>>load balancing and fault-tolerance. Anyone?
>>>
>>>Alex
>>>
>>>On Thu, 9 May 2002, Min Lu wrote:
>>>
>>>
>>>>Has anyone successfully built Wackamole on Solaris 8?
>>>>I just spent two days trying to get it to compile in
>>>>vain. There is no documentation. Where can I get the
>>>>binary of Wackamole if I just can't build it?
>>>>
>>>>Thanks.
>>>>
>>>>Min
>>>>

Jason> _________________________________________________________________
Jason> Chat with friends online, try MSN Messenger: http://messenger.msn.com

Jason> _______________________________________________
Jason> wackamole-users mailing list
Jason> wackamole-users [at] lists
Jason> http://lists.backhand.org/mailman/listinfo/wackamole-users

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads