Re: [Pacemaker] Pacemaker cluster: OpenAis communication channels
Steve, what has repeatedly come up is that RRP links don't auto-heal (see thread: http://oss.clusterlabs.org/pipermail/pacemaker/2009-May/001784.html), and that passive mode RRP seems to not work at all (see thread: https://lists.linux-foundation.org/pipermail/openais/2009-October/013095.html -- this was also heavily discussed on IRC; the only approach that fixed the issue was to change rrp_mode to active). Can you fill us in on the progress on these issues? Thanks! Cheers, Florian On 10/22/2009 06:14 AM, Steven Dake wrote: You can run with one NIC (and switch) but then your NIC and switch become a SPOF (single point of failure). Vehicles have a spare tire for a reason :) If a NIC fails it may be ok to switch a service to a different node. If a switch fails, The entire cluster becomes disabled until the switch returns to operation. Availability is a mathematical equation: A = MTTF / (MTTF+MTTR) Pacemaker improves availability (A) by reducing mean time to repair (MTTR) using failover while keeping the mean time to failure (MTTF) essentially the same (although it is generally a bit lower because of other components in the system required to introduce redundancy). Instead of a typical 1 machine MTTR of 4 hours under a typical SLA, MTTR may be 5-10 seconds or less (the time to failover the application and restart it). If MTTR is several days to service a switch, your availability may not meet your customer SLA obligations. When determining whether to use a redundant switch the risks vs cost have to be evaluated based upon your availability requirements. Regards -steve signature.asc Description: OpenPGP digital signature ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] Pacemaker cluster: OpenAis communication channels
Florian, I have checked the different links you have sent and I am a bit confused with this RRP matter. I want to use two machines (primary / secondary) and set up the following interfaces: - eth0 : direct link between primary and secondary - bond0: bonding of eth1 and eth2 (redundant network with two switches). Will there be any issues while using these two interfaces for OpenAis communication channels? (especially with the bonding) Thank you. 2009/10/22 Florian Haas florian.h...@linbit.com Steve, what has repeatedly come up is that RRP links don't auto-heal (see thread: http://oss.clusterlabs.org/pipermail/pacemaker/2009-May/001784.html), and that passive mode RRP seems to not work at all (see thread: https://lists.linux-foundation.org/pipermail/openais/2009-October/013095.html -- this was also heavily discussed on IRC; the only approach that fixed the issue was to change rrp_mode to active). Can you fill us in on the progress on these issues? Thanks! Cheers, Florian On 10/22/2009 06:14 AM, Steven Dake wrote: You can run with one NIC (and switch) but then your NIC and switch become a SPOF (single point of failure). Vehicles have a spare tire for a reason :) If a NIC fails it may be ok to switch a service to a different node. If a switch fails, The entire cluster becomes disabled until the switch returns to operation. Availability is a mathematical equation: A = MTTF / (MTTF+MTTR) Pacemaker improves availability (A) by reducing mean time to repair (MTTR) using failover while keeping the mean time to failure (MTTF) essentially the same (although it is generally a bit lower because of other components in the system required to introduce redundancy). Instead of a typical 1 machine MTTR of 4 hours under a typical SLA, MTTR may be 5-10 seconds or less (the time to failover the application and restart it). If MTTR is several days to service a switch, your availability may not meet your customer SLA obligations. When determining whether to use a redundant switch the risks vs cost have to be evaluated based upon your availability requirements. Regards -steve ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] Load Balancing, Node Scores and Stickiness
On Wed, Oct 21, 2009 at 2:06 PM, Colin colin@gmail.com wrote: Hi All, it seems from the documentation that Pacemaker has some inherent tendency to to load-balancing, in the sense of, given equal choice, not starting all resources on a single node... Now, I would like to be able to choose freely on a scale between always move everything to achieve good load balancing and don't gratuitously migrate resources, and would therefore like to understand the algorithms in Pacemaker better. Given a bunch of nodes and resources with a simple setup, i.e. no resource colocation constraints, no groups etc., I understand that a global score is calculated for each resource and each node, where score( resource, node ) = sum of all rsc_location constraints for that resource and node + if the resource is already running on this node, the stickiness (the stickiness of the resource or the global default stickiness) How does the assignment of nodes proceed? My guess is something like: for every resource in order of resource priority choose node with highest score for that resource if multiple nodes exist with the same score, pick one with the least allocated resources ??? somehow modify scores to prevent all resources on one node Are the details documented anywhere [except for the source]? Just the source so far ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] Load Balancing, Node Scores and Stickiness
On Thu, Oct 22, 2009 at 1:58 PM, Johan Verrept johan.verr...@able.be wrote: On Thu, 2009-10-22 at 12:37 +0200, Andrew Beekhof wrote: That's easy enough to understand ... and I can't do any fine-tuning, i.e. suppose that 4 nodes of my 10 node cluster fail, and then come up again. If all resources have equal score on all nodes (without counting stickiness), then (a) if stickiness is greater than 0 all resources will stay put, and (b) if stickiness is 0 then the cluster will move around resources to distribute them evenly? yep I wondered, does it happen dynamically? If one resource starts using a lot of resources, are the other migrated to other nodes? Not yet. Such a feature is planned though. At the moment pacemaker purely goes on the number of services it has allocated to the node. Total/Available RAM, CPU, HDD, none of these things are yet taken into account. ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
[Pacemaker] how to run a remote script
Hi I'm a newbie of Pacemaker and I'm working for Funambol company (www.funambol.com). I'm evaluating how to support a cluster solution for our product. I understood how to create a my own configuration with some monitoring commands. I found it in the main documentation, for instance for apache service now, I installed a Apache-Tomcat web container on the 2 nodes and my purpose is to start/stop these services I don't understand if I can set this information in the configuration of Pacemaker. could anybody give me some hints? thanks in advance gilberto ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] how to run a remote script
Hi Schmidt I'll try, thanks for the info gilberto Schmidt, Torsten wrote: Hi Gilberto, take a look at the tomcat ocf-script: start 'crm' to to 'ra' and do: crm(live)ra# meta tomcat ocf here you see the parameters (required and optional) implement the tomcat as primitive resource dependend on a cluster-ip (IPaddr2) like this: primitive res.ocf.tomct ocf:heartbeat:tomcat \ params name=value ...\ op monitor interval=30s timeout=10s primitive res.ip.tomcat ocf:heartbeat:IPaddr2 \ params ip=ip-address nic=eth0 cidr_netmask=22 \ op monitor interval=2s timeout=1s then bind this two together as a colocation and tell pacemaker the order, how to start + stop: colocation co.res.ocf.tomcat_on_res.ip.tomcat inf: res.ocf.tomcat res.ip.tomcat order o.res.ip.tomcat_before_res.ocf.tomcat inf: res.ip.tomcat res.ocf.tomcat:start and look at the documentation on http://clusterlabs.org/wiki/Documentation Mit freundlichen Grüßen / with kind regards Torsten Schmidt -Ursprüngliche Nachricht- Von: gilberto migliavacca [mailto:gbmig...@yahoo.it] Gesendet: Donnerstag, 22. Oktober 2009 15:55 An: pacemaker@oss.clusterlabs.org Betreff: [Pacemaker] how to run a remote script Hi I'm a newbie of Pacemaker and I'm working for Funambol company (www.funambol.com). I'm evaluating how to support a cluster solution for our product. I understood how to create a my own configuration with some monitoring commands. I found it in the main documentation, for instance for apache service now, I installed a Apache-Tomcat web container on the 2 nodes and my purpose is to start/stop these services I don't understand if I can set this information in the configuration of Pacemaker. could anybody give me some hints? thanks in advance gilberto ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] how to run a remote script
Hi I need a step behind now I installed a java stand alone process with its onw start and stop scripts on the 2 nodes and my purpose is to start/stop these services using pacemaker how can I create the myapp resource agent on the nodes, somthing like the ocf:heartbeat:tomcat? thanks in advance gilberto Schmidt, Torsten wrote: Hi Gilberto, take a look at the tomcat ocf-script: start 'crm' to to 'ra' and do: crm(live)ra# meta tomcat ocf here you see the parameters (required and optional) implement the tomcat as primitive resource dependend on a cluster-ip (IPaddr2) like this: primitive res.ocf.tomct ocf:heartbeat:tomcat \ params name=value ...\ op monitor interval=30s timeout=10s primitive res.ip.tomcat ocf:heartbeat:IPaddr2 \ params ip=ip-address nic=eth0 cidr_netmask=22 \ op monitor interval=2s timeout=1s then bind this two together as a colocation and tell pacemaker the order, how to start + stop: colocation co.res.ocf.tomcat_on_res.ip.tomcat inf: res.ocf.tomcat res.ip.tomcat order o.res.ip.tomcat_before_res.ocf.tomcat inf: res.ip.tomcat res.ocf.tomcat:start and look at the documentation on http://clusterlabs.org/wiki/Documentation Mit freundlichen Grüßen / with kind regards Torsten Schmidt -Ursprüngliche Nachricht- Von: gilberto migliavacca [mailto:gbmig...@yahoo.it] Gesendet: Donnerstag, 22. Oktober 2009 15:55 An: pacemaker@oss.clusterlabs.org Betreff: [Pacemaker] how to run a remote script Hi I'm a newbie of Pacemaker and I'm working for Funambol company (www.funambol.com). I'm evaluating how to support a cluster solution for our product. I understood how to create a my own configuration with some monitoring commands. I found it in the main documentation, for instance for apache service now, I installed a Apache-Tomcat web container on the 2 nodes and my purpose is to start/stop these services I don't understand if I can set this information in the configuration of Pacemaker. could anybody give me some hints? thanks in advance gilberto ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
[Pacemaker] Can I edit apache configuration file using pacemaker?
Hi I have this problem and I'd like to know if I can solve it using pacemaker I have 2 nodes with apache + modjk + tomcat on both nodes, moreover the modjk components point to the tomcat on both nodes in this way the configuration (worker.properties) file of the modjk keeps the information about the tomcat on the 2 nodes. something like: apache apache modjk modjk \ / \ / tomcat-- / \-- tomcat I'd like to understand if I can modify the modjk configuration file (worker.properties) using some pacemaker commands in this way I can - deactivate a node (modifying the modjk and launching a graceful command to apache - wait until the node finishes up all the active session - apply a patch on the node - restart the node all the steps about start/stop I understood I can do it with pacemaker, my problem is about the file modification. Can I edit the modjk configuration file (worker.properties) using some pacemaker commands? thanks in advance ... sorry for this complicated question regards gilberto ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Re: [Pacemaker] Can I edit apache configuration file using pacemaker?
Hi, On Thu, Oct 22, 2009 at 07:25:27PM +0200, gilberto migliavacca wrote: Hi I have this problem and I'd like to know if I can solve it using pacemaker I have 2 nodes with apache + modjk + tomcat on both nodes, moreover the modjk components point to the tomcat on both nodes in this way the configuration (worker.properties) file of the modjk keeps the information about the tomcat on the 2 nodes. something like: apache apache modjk modjk \ / \ / tomcat-- / \-- tomcat I'd like to understand if I can modify the modjk configuration file (worker.properties) using some pacemaker commands in this way I can - deactivate a node (modifying the modjk and launching a graceful command to apache - wait until the node finishes up all the active session - apply a patch on the node - restart the node all the steps about start/stop I understood I can do it with pacemaker, my problem is about the file modification. Can I edit the modjk configuration file (worker.properties) using some pacemaker commands? thanks in advance ... sorry for this complicated question All the cluster knows is how to start, stop, and monitor a resource. The resource is represented and directly controlled by the corresponding resource agent. If you can fit whatever you want to do into a script and then have it do thing 1 on start and thing 2 on stop and that it returns a proper status when invoked with the monitor action (though there are also scripts which are stateless, take a look for example at ocf:heartbeat:Dummy), then you should be good to go. But note that one resource should not influence the status of another resource while it's running. Thanks, Dejan regards gilberto ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker ___ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker