Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Linux-HA: Users

active/active

  

 
Linux-HA users RSS feed   Index | Next | Previous | View Threaded


qgriffith22 at yahoo

Jun 16, 2003, 8:18 AM

Post #1 of 13 (437 views)
Permalink
active/active
Does heartbeat offer the ability to have an active
active setup where either box will failover to each
other along with the apps that live on that box. So
if nodea is running Apache, and nodeb is running
mysql, if either node goes down I want the app to go
each machine. The haresource script is suppose to be
the same on each box so I do not know how this could
be done.

__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com


alanr at unix

Jun 16, 2003, 8:33 AM

Post #2 of 13 (430 views)
Permalink
active/active [In reply to]
Quenten Griffith wrote:
> Does heartbeat offer the ability to have an active
> active setup where either box will failover to each
> other along with the apps that live on that box. So
> if nodea is running Apache, and nodeb is running
> mysql, if either node goes down I want the app to go
> each machine. The haresource script is suppose to be
> the same on each box so I do not know how this could
> be done.

haresource file:
nodea IPADDR::ip-address-for-apache apache
nodeb IPADDR::ip-address-for-mysql mysql

This sounds like something to add to the FAQ...

--
Alan Robertson <alanr [at] unix>

"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce


qgriffith22 at yahoo

Jun 16, 2003, 9:30 AM

Post #3 of 13 (431 views)
Permalink
active/active [In reply to]
Something so easy and I missed it, thank you.
--- Alan Robertson <alanr [at] unix> wrote:
> Quenten Griffith wrote:
> > Does heartbeat offer the ability to have an active
> > active setup where either box will failover to
> each
> > other along with the apps that live on that box.
> So
> > if nodea is running Apache, and nodeb is running
> > mysql, if either node goes down I want the app to
> go
> > each machine. The haresource script is suppose to
> be
> > the same on each box so I do not know how this
> could
> > be done.
>
> haresource file:
> nodea IPADDR::ip-address-for-apache apache
> nodeb IPADDR::ip-address-for-mysql mysql
>
> This sounds like something to add to the FAQ...
>
> --
> Alan Robertson <alanr [at] unix>
>
> "Openness is the foundation and preservative of
> friendship... Let me claim
> from you at all times your undisguised opinions." -
> William Wilberforce
>


__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com


alanr at unix

Jun 16, 2003, 9:56 AM

Post #4 of 13 (433 views)
Permalink
active/active [In reply to]
Quenten Griffith wrote:
> Something so easy and I missed it, thank you.


Don't feel bad. Lots of people miss this. It seems so natural to me, but
obviously not to everyone. Check out the new version of the faqntips.html
document here:

http://linux-ha.org/download/faqntips.html

The answer is here:
http://linux-ha.org/download/faqntips.html#active_active

Let me know if that answers your question sufficiently.

Thanks!

--
Alan Robertson <alanr [at] unix>

"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce


qgriffith22 at yahoo

Jun 16, 2003, 10:22 AM

Post #5 of 13 (431 views)
Permalink
active/active [In reply to]
Yes it does and thank you. Also learned from reading
it that it will prevent IP takeover (but is being
worked on) That is something I will need to have is
IPFAIL or have two DNS entries so that it will point
to both boxes just have one ranked higher then the
other. Or I can try to figure a way out to right
inside the app start stop script that haresource calls
it will bring down eth0:0 and bring it up on the other
node.
--- Alan Robertson <alanr [at] unix> wrote:
> Quenten Griffith wrote:
> > Something so easy and I missed it, thank you.
>
>
> Don't feel bad. Lots of people miss this. It seems
> so natural to me, but
> obviously not to everyone. Check out the new
> version of the faqntips.html
> document here:
>
> http://linux-ha.org/download/faqntips.html
>
> The answer is here:
>
>
http://linux-ha.org/download/faqntips.html#active_active
>
> Let me know if that answers your question
> sufficiently.
>
> Thanks!
>
> --
> Alan Robertson <alanr [at] unix>
>
> "Openness is the foundation and preservative of
> friendship... Let me claim
> from you at all times your undisguised opinions." -
> William Wilberforce
>


__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com


alanr at unix

Jun 16, 2003, 1:10 PM

Post #6 of 13 (440 views)
Permalink
active/active [In reply to]
Quenten Griffith wrote:
> Yes it does and thank you. Also learned from reading
> it that it will prevent IP takeover (but is being
> worked on) That is something I will need to have is
> IPFAIL or have two DNS entries so that it will point
> to both boxes just have one ranked higher then the
> other. Or I can try to figure a way out to right
> inside the app start stop script that haresource calls
> it will bring down eth0:0 and bring it up on the other
> node.
> --- Alan Robertson <alanr [at] unix> wrote:


Ipfail isn't just about IP failover. Heartbeat will fail over IP addresses
- it just won't automatically move resources around to make sure they're on
the right machine if you lose important internet connectivity.


--
Alan Robertson <alanr [at] unix>

"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce


qgriffith22 at yahoo

Jun 16, 2003, 2:18 PM

Post #7 of 13 (431 views)
Permalink
active/active [In reply to]
How soon do you think that the code will be there to
have ipfail and active/active support? What are the
current limitations. I am coming for an AIX failover
(HACMP) background and that is what I enjoy about
Heartbeat is it is easy to maintain and pretty easy to
setup and uses basically three conf files sort of like
HACMP uses a few conf files and scripts to work. I am
pushing heartbeat hard at my work. They have some
product called SteelEye which is a non free product
they want to use for Failover. I of course would
rather user the Open Source failover that comes with
almost all the distros of linux then a third party
program. Right now there is now plan on doing an
active/active setup with Linux, I just want to be
prepared when or if we do, and am glad to help any way
I can with heartbeat.
--- Alan Robertson <alanr [at] unix> wrote:
> Quenten Griffith wrote:
> > Yes it does and thank you. Also learned from
> reading
> > it that it will prevent IP takeover (but is being
> > worked on) That is something I will need to have
> is
> > IPFAIL or have two DNS entries so that it will
> point
> > to both boxes just have one ranked higher then the
> > other. Or I can try to figure a way out to right
> > inside the app start stop script that haresource
> calls
> > it will bring down eth0:0 and bring it up on the
> other
> > node.
> > --- Alan Robertson <alanr [at] unix> wrote:
>
>
> Ipfail isn't just about IP failover. Heartbeat will
> fail over IP addresses
> - it just won't automatically move resources around
> to make sure they're on
> the right machine if you lose important internet
> connectivity.
>
>
> --
> Alan Robertson <alanr [at] unix>
>
> "Openness is the foundation and preservative of
> friendship... Let me claim
> from you at all times your undisguised opinions." -
> William Wilberforce
>


__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com


alanr at unix

Jun 16, 2003, 2:28 PM

Post #8 of 13 (430 views)
Permalink
active/active [In reply to]
Quenten Griffith wrote:
> How soon do you think that the code will be there to
> have ipfail and active/active support?

This is pretty close to being there in CVS now. The heartbeat part is done,
but the ipfail part isn't yet done. But, I don't think it's that difficult.
Kevin simply hasn't had time to look at it yet. Ipfail is < 600 lines of
code or so if you want to look at it. Basically, the idea is that if the
two nodes have voted before and moved resources to one machine or another,
then the machine which has no resources needs to send a message which is
almost exactly like the one it's sending now. But, it would have to know
that auto_failback (the new option which replaces nice_failback) is enabled...

> What are the
> current limitations. I am coming for an AIX failover
> (HACMP) background and that is what I enjoy about
> Heartbeat is it is easy to maintain and pretty easy to
> setup and uses basically three conf files sort of like
> HACMP uses a few conf files and scripts to work. I am
> pushing heartbeat hard at my work. They have some
> product called SteelEye which is a non free product
> they want to use for Failover. I of course would
> rather user the Open Source failover that comes with
> almost all the distros of linux then a third party
> program. Right now there is now plan on doing an
> active/active setup with Linux, I just want to be
> prepared when or if we do, and am glad to help any way
> I can with heartbeat.

Of course, it supports active/active now. However, the nice capabilities
which ipfail brings to the table are not available if you run active/active.
We're always happy to have people send us nice patches.

--
Alan Robertson <alanr [at] unix>

"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce


lmb at suse

Jun 17, 2003, 8:23 AM

Post #9 of 13 (430 views)
Permalink
active/active [In reply to]
On 2003-06-16T14:18:53,
Quenten Griffith <qgriffith22 [at] yahoo> said:

> pushing heartbeat hard at my work. They have some
> product called SteelEye which is a non free product
> they want to use for Failover. I of course would
> rather user the Open Source failover that comes with
> almost all the distros of linux then a third party
> program.

Truth be told, SteelEye LifeKeeper is a lot more powerful than heartbeat
- >2 nodes, service monitoring etc. If you need any of these features,
it is a good choice right now.


Sincerely,
Lars Marowsky-Brée <lmb [at] suse>

--
SuSE Labs - Research & Development, SuSE Linux AG

"If anything can go wrong, it will." "Chance favors the prepared (mind)."
-- Capt. Edward A. Murphy -- Louis Pasteur


linux-ha at lists

Jul 23, 2003, 2:31 PM

Post #10 of 13 (435 views)
Permalink
Re: active/active [In reply to]
Digging up an old thread,

Quoting Alan Robertson <alanr [at] unix>:

> This is pretty close to being there in CVS now. The heartbeat part is done,
> but the ipfail part isn't yet done. But, I don't think it's that difficult.

Quenten, I sent a patch to the linux-ha-dev list a few days ago which you can
retrieve here: http://marc.theaimsgroup.com/?l=linux-ha-dev&m=105847642325431&q=p5
(Side note: I mislabled the MIME type, but rest assured that it is text/plain)

This only applies to the development code, and I realize that this is the user
(i.e.: non-devel) list, but if you have the chance, I'd love to get your
feedback on whether it works for you. This is also a request for other
potential testers who need active/active functionality in combination with ipfail.

In order to use it, you should check out a copy of the current CVS code and
apply the patch with the 'patch' command. Something like:

patch -p0 < ipfail-auto_failback-1.patch

from the linux-ha/contrib/ipfail/ directory would do it I think. I haven't
tried, but this should also apply cleanly to the beta 1.1.2 release, as I don't
think the ipfail code has changed in CVS since then.

Feedback is very much welcomed. Thanks.


bruno.macadre at univ-rouen

Aug 2, 2012, 12:37 AM

Post #11 of 13 (286 views)
Permalink
Re: Active/Active [In reply to]
Hi,

I'm working on a cluster of the same type as yours and I still ask
myself some questions. Every feedback will be appreciated :

- What OS did you use ?
I do a lot of work on Ubuntu 12.04 server but I'm not satisfied by the
cluster stability mainly with OCFS2

- Do you use clustered filesystem (like OCFS2 or GFS2) in you Dual
Primary DRBD ? Did you have special concerns ?
I had a lot of concern for use OCFS2 on Ubuntu server 'cause it need a
cman stack above corosync which made entire cluster unstable. For
example : When I do 'node standby node1' it kill pacemaker whithout the
possibility to comme back online at all !!!

- Have you been encountering some obstacles when NFS resources switch
during high use ?

I gladly give a look at your cib.xml file if you wish.

Regards,
Bruno

Le 01/08/2012 11:22, Yount, William D a écrit :
> I was wondering if someone could give a look over my cib.xml file and see if I need to change anything. I am attempting to create an Active/Active cluster offering up a DRBD volume for NFS share. Everything works fine as it is, but I would like someone more knowledgeable to look it over at your leisure.
>
>
>
> Thank you,
> william
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>


--

Bruno MACADRE
-------------------------------------------------------------------
Ingénieur Systèmes et Réseau | Systems and Network Engineer
Département Informatique | Department of computer science
Responsable Réseau et Téléphonie | Telecom and Network Manager
Université de Rouen | University of Rouen
-------------------------------------------------------------------
Coordonnées / Contact :
Université de Rouen
Faculté des Sciences et Techniques - Madrillet
Avenue de l'Université - BP12
76801 St Etienne du Rouvray CEDEX
FRANCE

Tél : +33 (0)2-32-95-51-86
Fax : +33 (0)2-32-95-51-87
-------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems


Yount.William at menloworldwide

Aug 2, 2012, 1:22 AM

Post #12 of 13 (287 views)
Permalink
Re: Active/Active [In reply to]
I am using CentOS 6.2, mainly because that is that is what the "Clusters from Scratch" documentation uses.
I am using GFS2.

Yeah, I had some issues with the cman/corosync/pacemaker startup order at first. But CentOS uses chkconfig which allowed me to turn corosync off, cman on and pacemaker on. Ever since, I haven't had an issue with the cluster coming up on its own. Previously I had to stop coryosync, start cman and then start pacemaker.

I do seem to have some issues with the cluster recognizing when a resource is down but I believe that may be tied to pacemaker which I am using for fencing. There is something wrong with my cib.xml configuration. All the resources seem to be "docked" at Node2. If I cut Node2 off or unplug a network cable from Node2 than my cluster goes down. Even my cloned IP address goes down.


-----Original Message-----
From: linux-ha-bounces [at] lists [mailto:linux-ha-bounces [at] lists] On Behalf Of Bruno MACADRE
Sent: Thursday, August 02, 2012 2:38 AM
To: linux-ha [at] lists
Subject: Re: [Linux-HA] Active/Active

Hi,

I'm working on a cluster of the same type as yours and I still ask myself some questions. Every feedback will be appreciated :

- What OS did you use ?
I do a lot of work on Ubuntu 12.04 server but I'm not satisfied by the cluster stability mainly with OCFS2

- Do you use clustered filesystem (like OCFS2 or GFS2) in you Dual Primary DRBD ? Did you have special concerns ?
I had a lot of concern for use OCFS2 on Ubuntu server 'cause it need a cman stack above corosync which made entire cluster unstable. For example : When I do 'node standby node1' it kill pacemaker whithout the possibility to comme back online at all !!!

- Have you been encountering some obstacles when NFS resources switch during high use ?

I gladly give a look at your cib.xml file if you wish.

Regards,
Bruno

Le 01/08/2012 11:22, Yount, William D a écrit :
> I was wondering if someone could give a look over my cib.xml file and see if I need to change anything. I am attempting to create an Active/Active cluster offering up a DRBD volume for NFS share. Everything works fine as it is, but I would like someone more knowledgeable to look it over at your leisure.
>
>
>
> Thank you,
> william
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>


--

Bruno MACADRE
-------------------------------------------------------------------
Ingénieur Systèmes et Réseau | Systems and Network Engineer
Département Informatique | Department of computer science
Responsable Réseau et Téléphonie | Telecom and Network Manager
Université de Rouen | University of Rouen
-------------------------------------------------------------------
Coordonnées / Contact :
Université de Rouen
Faculté des Sciences et Techniques - Madrillet
Avenue de l'Université - BP12
76801 St Etienne du Rouvray CEDEX
FRANCE

Tél : +33 (0)2-32-95-51-86
Fax : +33 (0)2-32-95-51-87
-------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems


david at davidcoulson

Aug 2, 2012, 4:24 AM

Post #13 of 13 (286 views)
Permalink
Re: Active/Active [In reply to]
From your post on the drbd list it looks like your storage is messed up.

Post the output of crm_mon -1fr from both nodes. Guessing your drbd resource didn't promote to master on one node because your drbd cluster is out of sync.

Sent from my iPad

On Aug 2, 2012, at 4:22 AM, "Yount, William D" <Yount.William [at] menloworldwide> wrote:

> I am using CentOS 6.2, mainly because that is that is what the "Clusters from Scratch" documentation uses.
> I am using GFS2.
>
> Yeah, I had some issues with the cman/corosync/pacemaker startup order at first. But CentOS uses chkconfig which allowed me to turn corosync off, cman on and pacemaker on. Ever since, I haven't had an issue with the cluster coming up on its own. Previously I had to stop coryosync, start cman and then start pacemaker.
>
> I do seem to have some issues with the cluster recognizing when a resource is down but I believe that may be tied to pacemaker which I am using for fencing. There is something wrong with my cib.xml configuration. All the resources seem to be "docked" at Node2. If I cut Node2 off or unplug a network cable from Node2 than my cluster goes down. Even my cloned IP address goes down.
>
>
> -----Original Message-----
> From: linux-ha-bounces [at] lists [mailto:linux-ha-bounces [at] lists] On Behalf Of Bruno MACADRE
> Sent: Thursday, August 02, 2012 2:38 AM
> To: linux-ha [at] lists
> Subject: Re: [Linux-HA] Active/Active
>
> Hi,
>
> I'm working on a cluster of the same type as yours and I still ask myself some questions. Every feedback will be appreciated :
>
> - What OS did you use ?
> I do a lot of work on Ubuntu 12.04 server but I'm not satisfied by the cluster stability mainly with OCFS2
>
> - Do you use clustered filesystem (like OCFS2 or GFS2) in you Dual Primary DRBD ? Did you have special concerns ?
> I had a lot of concern for use OCFS2 on Ubuntu server 'cause it need a cman stack above corosync which made entire cluster unstable. For example : When I do 'node standby node1' it kill pacemaker whithout the possibility to comme back online at all !!!
>
> - Have you been encountering some obstacles when NFS resources switch during high use ?
>
> I gladly give a look at your cib.xml file if you wish.
>
> Regards,
> Bruno
>
> Le 01/08/2012 11:22, Yount, William D a écrit :
>> I was wondering if someone could give a look over my cib.xml file and see if I need to change anything. I am attempting to create an Active/Active cluster offering up a DRBD volume for NFS share. Everything works fine as it is, but I would like someone more knowledgeable to look it over at your leisure.
>>
>>
>>
>> Thank you,
>> william
>> _______________________________________________
>> Linux-HA mailing list
>> Linux-HA [at] lists
>> http://lists.linux-ha.org/mailman/listinfo/linux-ha
>> See also: http://linux-ha.org/ReportingProblems
>>
>
>
> --
>
> Bruno MACADRE
> -------------------------------------------------------------------
> Ingénieur Systèmes et Réseau | Systems and Network Engineer
> Département Informatique | Department of computer science
> Responsable Réseau et Téléphonie | Telecom and Network Manager
> Université de Rouen | University of Rouen
> -------------------------------------------------------------------
> Coordonnées / Contact :
> Université de Rouen
> Faculté des Sciences et Techniques - Madrillet
> Avenue de l'Université - BP12
> 76801 St Etienne du Rouvray CEDEX
> FRANCE
>
> Tél : +33 (0)2-32-95-51-86
> Fax : +33 (0)2-32-95-51-87
> -------------------------------------------------------------------
>
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Linux-HA users RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.