[opensuse] cluster solutions ?
Hello Is there some "cluster solutions" with opensuse ? any infos, links welcome thanks a lot -- Cordialement Frank Bonnet -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Frank Bonnet wrote:
Hello
Is there some "cluster solutions" with opensuse ? any infos, links welcome
Clustering just makes use of the normally unused "a" "b" and "c" runlevels. a "heartbeat" program is used to monitor the other cluster members, and then used to take appropriate action to start up or shut down processes in response to which other host(s) in the cluster are up or down. man init to read about run levels a b and c. More sophisticated clustering software basically just implements a state machine. for example, if you have a three-host cluster, then the valid states are: Host3 Host2 Host1 Cluster-state down down up 1 down up down 2 up down up 3 down up up 12 up up down 23 up down up 13 up up up 123 In configuring each host, you just determine, for each of the above states, what processes should be running on each host (for example, if host 2 is down, some of the programs that host 2 hormally runs should be started on host 1, some maybe on host 3, and a remainder (or none) are just unavailable until host 2 is back up. The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software. For a two-host cluster, you can completely implement the cluster with runlevels a, b, and c. Run level a for normal cluster operations. Run level b for when the other host is down. Run level c for deliberately leaving the cluster. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Aaron Kulkis wrote:
The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software.
There is software for doing high availability clusters in the suse repos, it's called heartbeat. The heartbeat website has tutorials. It does have a learning curve, and some assembly required. There are some tutorials on the site - http://linux-ha.org/ Joe -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Or, does the question concern HPC Clusters, which are totally/completely/thoroughly different from HA (or "Enterprise") clusters? If so, my vote is to keep it simple with tools like YACI/pdsh/conman from LLNL: http://www.llnl.gov/linux/ On Nov 27, 2007 7:24 PM, Joe Sloan <joe@tmsusa.com> wrote:
Aaron Kulkis wrote:
The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software.
There is software for doing high availability clusters in the suse repos, it's called heartbeat. The heartbeat website has tutorials. It does have a learning curve, and some assembly required.
There are some tutorials on the site -
Joe
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Joe Sloan wrote:
Aaron Kulkis wrote:
The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software.
There is software for doing high availability clusters in the suse repos, it's called heartbeat. The heartbeat website has tutorials. It does have a learning curve, and some assembly required.
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1. My personal experience with HA systems have been a Sequent Cluster I used to admin in the 90's, and indirect dealings as assistant supervisor at the General Motors Supercomputing site for about a year in 2001/2002 (then GM layed me off, then the US Army mobilized me).
There are some tutorials on the site -
Thanks for the URL.
Joe
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1.
Its in 10.2 and 10.3. IIRC heartbeat had some issues with the 2.6.x kernels early on, so it may have been dropped from the distro until that was resolved. FYI: Novell is a major sponsor of heartbeat. They have at least 2 paid developers and maybe more supporting it as a primary part of their job. And it is officially supported in SLES. If you want the latest and greatest, they have interim package releases available in the build service. I think the interims are designed for QA testing, not necessarily production use, but they still get a lot of Novell internal testing prior to being released on the build service. Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Nov 29, 2007, at 6:27 PM, Greg Freemyer wrote:
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1.
Its in 10.2 and 10.3. IIRC heartbeat had some issues with the 2.6.x kernels early on, so it may have been dropped from the distro until that was resolved.
the only thing i can think of that might affect heartbeat is a scheduler bug - but AFAIK that only happened in one specific RedHat kernel. no idea why it got left out of 10.1
FYI: Novell is a major sponsor of heartbeat. They have at least 2 paid developers and maybe more supporting it as a primary part of their job. And it is officially supported in SLES. If you want the latest and greatest, they have interim package releases available in the build service. I think the interims are designed for QA testing, not necessarily production use,
they're intended to be suitable for production
but they still get a lot of Novell internal testing prior to being released on the build service.
Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf
The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2007-12-01T14:25:35, Andrew Beekhof <beekhof@gmail.com> wrote:
the only thing i can think of that might affect heartbeat is a scheduler bug - but AFAIK that only happened in one specific RedHat kernel.
no idea why it got left out of 10.1
It wasn't, it was in 10.1. -- Teamlead Kernel, SuSE Labs, Research and Development SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) "Experience is the name everyone gives to their mistakes." -- Oscar Wilde -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wed, 2007-11-28 at 07:17 -0500, Aaron Kulkis wrote:
Joe Sloan wrote:
Aaron Kulkis wrote:
The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software.
There is software for doing high availability clusters in the suse repos, it's called heartbeat. The heartbeat website has tutorials. It does have a learning curve, and some assembly required.
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1.
look at: http://ftp5.gwdg.de/pub/opensuse/repositories/server:/ha-clustering/openSUSE... sle, open 10.0 - 10.3, fedora, ubuntu builds, they're all there.. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 * Aaron Kulkis <akulkis00@hotpop.com> [01-01-70 12:34]: [...]
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1. [...]
09:50 wahoo:~ > webpin heartbeat Query URL: http://benjiweber.co.uk:8080/searchservice/SearchService/Search/Simple/SUSE_... 22 results (12 packages) found for "heartbeat" in SUSE_Linux_101 * heartbeat: The Heartbeat Subsystem for High-Availability Linux - 2.0.7 [suse-update(S:y)] {i586} @ http://ftp.gwdg.de/pub/suse/update/10.1 - 2.0.5 [BS::SUSE:/SL-10.1(S:n)] {ppc} @ http://download.opensuse.org/repositories/SUSE:/SL-10.1/standard - 2.1.2 [BS::server:/ha-clustering(S:n)] {src} [...] @ * heartbeat-stonith: An Interface for Shoot The Other Node In The Head - 2.0.7 [suse-update(S:y)] {i586} @ http://ftp.gwdg.de/pub/suse/update/10.1 - 2.0.5 [BS::SUSE:/SL-10.1(S:n)] {ppc} @ http://download.opensuse.org/repositories/SUSE:/SL-10.1/standard - 2.1.2 [BS::server:/ha-clustering(S:n)] {i586} @ http://download.opensuse.org/repositories/server:/ha-clustering/SUSE_Linux_1... - -- Patrick Shanahan Plainfield, Indiana, USA HOG # US1244711 http://wahoo.no-ip.org Photo Album: http://wahoo.no-ip.org/gallery2 Registered Linux User #207535 @ http://counter.li.org -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) iD8DBQFHUXUsClSjbQz1U5oRAk8zAJ4gta1UMm14YesJQ6+o8Yl3yMeGuQCeL7ub xGLa7tG+zYOdaNCVr6qr3ZM= =4gEL -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Patrick Shanahan wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
* Aaron Kulkis <akulkis00@hotpop.com> [01-01-70 12:34]: [...]
Back in the 8.x and 9.x, heartbeat and stonith (shoot the other node in the head) were part of the all of the "SusE _._ Professional" distros. I was disappointed to not see it in 10.1. [...]
09:50 wahoo:~ > webpin heartbeat Query URL: http://benjiweber.co.uk:8080/searchservice/SearchService/Search/Simple/SUSE_... 22 results (12 packages) found for "heartbeat" in SUSE_Linux_101 * heartbeat: The Heartbeat Subsystem for High-Availability Linux - 2.0.7 [suse-update(S:y)] {i586} @ http://ftp.gwdg.de/pub/suse/update/10.1 - 2.0.5 [BS::SUSE:/SL-10.1(S:n)] {ppc} @ http://download.opensuse.org/repositories/SUSE:/SL-10.1/standard - 2.1.2 [BS::server:/ha-clustering(S:n)] {src} [...] @ * heartbeat-stonith: An Interface for Shoot The Other Node In The Head - 2.0.7 [suse-update(S:y)] {i586} @ http://ftp.gwdg.de/pub/suse/update/10.1 - 2.0.5 [BS::SUSE:/SL-10.1(S:n)] {ppc} @ http://download.opensuse.org/repositories/SUSE:/SL-10.1/standard - 2.1.2 [BS::server:/ha-clustering(S:n)] {i586} @ http://download.opensuse.org/repositories/server:/ha-clustering/SUSE_Linux_1...
Strange that I didn't see it on the install DVD. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Nov 27, 2007, at 12:35 PM, Aaron Kulkis wrote:
Frank Bonnet wrote:
Hello Is there some "cluster solutions" with opensuse ? any infos, links welcome
Clustering just makes use of the normally unused "a" "b" and "c" runlevels.
a "heartbeat" program is used to monitor the other cluster members, and then used to take appropriate action to start up or shut down processes in response to which other host(s) in the cluster are up or down.
newer versions of heartbeat will also recover individual services on an otherwise healthy node. it also comes with a sophisticated policy engine so that you can define the order services need to start in, their preferred locations and which services must/must not run together.
man init
to read about run levels a b and c.
More sophisticated clustering software basically just implements a state machine. for example, if you have a three-host cluster, then the valid states are:
Host3 Host2 Host1 Cluster-state down down up 1 down up down 2 up down up 3 down up up 12 up up down 23 up down up 13 up up up 123
In configuring each host, you just determine, for each of the above states, what processes should be running on each host (for example, if host 2 is down, some of the programs that host 2 hormally runs should be started on host 1, some maybe on host 3, and a remainder (or none) are just unavailable until host 2 is back up.
The whole idea is VERY similar to the runlevel concept, except that you can (should) have a lot more states available, depending on the software.
For a two-host cluster, you can completely implement the cluster with runlevels a, b, and c.
Run level a for normal cluster operations. Run level b for when the other host is down. Run level c for deliberately leaving the cluster.
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2007-11-27T06:35:36, Aaron Kulkis <akulkis00@hotpop.com> wrote:
Clustering just makes use of the normally unused "a" "b" and "c" runlevels.
That's a very special example of clustering. I've never seen a cluster product which did; which one do you have in mind?
For a two-host cluster, you can completely implement the cluster with runlevels a, b, and c.
Run level a for normal cluster operations. Run level b for when the other host is down. Run level c for deliberately leaving the cluster.
Certainly an interesting approach, yes. Regards, Lars -- Teamlead Kernel, SuSE Labs, Research and Development SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) "Experience is the name everyone gives to their mistakes." -- Oscar Wilde -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Lars Marowsky-Bree wrote:
On 2007-11-27T06:35:36, Aaron Kulkis <akulkis00@hotpop.com> wrote:
Clustering just makes use of the normally unused "a" "b" and "c" runlevels.
That's a very special example of clustering. I've never seen a cluster product which did; which one do you have in mind?
I should have said that a very old method was by using the a, b, and c runlevels. Every clustering product I've ever seen has been an expansion of the regular run-levels, basically adding a "cluster-state" on top of the run-level state.
For a two-host cluster, you can completely implement the cluster with runlevels a, b, and c.
Run level a for normal cluster operations. Run level b for when the other host is down. Run level c for deliberately leaving the cluster.
Certainly an interesting approach, yes.
It's a very, very old method. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Lars Marowsky-Bree wrote:
On 2007-11-27T06:35:36, Aaron Kulkis <akulkis00@hotpop.com> wrote:
Clustering just makes use of the normally unused "a" "b" and "c" runlevels.
That's a very special example of clustering. I've never seen a cluster product which did; which one do you have in mind?
That's how it was done before specific clustering products were available... original clusters had only 2 nodes. this was back in the 80's. although I do NOT think that's what the a,b,c run levels were originally meant for...
For a two-host cluster, you can completely implement the cluster with runlevels a, b, and c.
Run level a for normal cluster operations. Run level b for when the other host is down. Run level c for deliberately leaving the cluster.
Certainly an interesting approach, yes.
Regards, Lars
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Sunday 09 December 2007 04:04:10 am Aaron Kulkis wrote:
although I do NOT think that's what the a,b,c run levels were originally meant for...
Where did you found reference to runlevels a,b,c in Linux? Text search that you use to gather information should be improved to refere only to Linux usage of the word runlevel. As it is now, this information is out of Linux context. -- Regards, Rajko -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Rajko M. wrote:
On Sunday 09 December 2007 04:04:10 am Aaron Kulkis wrote:
although I do NOT think that's what the a,b,c run levels were originally meant for...
Where did you found reference to runlevels a,b,c in Linux?
Oh, they've been around before Linux was around.... 4.3 BSD, I think man init
Text search that you use to gather information should be improved to refere only to Linux usage of the word runlevel. As it is now, this information is out of Linux context.
run level is the terminology which has been in use since at least the early 1980's.
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Nov 27, 2007 3:44 AM, Frank Bonnet <f.bonnet@esiee.fr> wrote:
Hello
Is there some "cluster solutions" with opensuse ? any infos, links welcome
What type of clustering? And what features do you need? For pure HA (like a webserver), try heartbeat (sponsored by Novell / IBM). For HA with redundant real-time data, try heartbeat + drbd Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tuesday 27 November 2007 09:44:46 Frank Bonnet wrote:
Hello
Is there some "cluster solutions" with opensuse ? any infos, links welcome
What type of cluster are you looking for? High Availability, or High Performance? Anders -- Madness takes its toll -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (11)
-
Aaron Kulkis -
Anders Johansson -
Andrew Beekhof -
Chris Worley -
Frank Bonnet -
Greg Freemyer -
Hans Witvliet -
Joe Sloan -
Lars Marowsky-Bree -
Patrick Shanahan -
Rajko M.