Just for informationin the past year, we've had ~4 customers who have required xCAT HA in their environments, and the systems were critical or big enough to account for that.
Below are the scenarios of what we have doneThe first cluster was a storage cluster managed by xCAT, and they had 2 sites, where they needed some kind of failover if one site went down.
The 2nd/3rd clusters were HPC clusters where we used drbd, and they worked really well; in both cases we had SGE, and in that case we used SGE's built in master/shadow to do the failover of the resource manager
lastly, the current project I am working on, between the 2 xCAT HA nodes, we have a shared storage, so in this case we are using GFS2 with clvmd, which so far has worked really well; and from the resource manager side, we have LSF, and have controlled that through cman/pacemaker making sure that the LSF_WORKDIR is mounted prior to starting of the lsf daemons. I may be documenting this for the wiki, and for public, once the project has been finalised.
I hope that answers your questions, given our use cases of HA requirementsNote on the HA capabilities of redhat, there were changes of the preferred stack for this, so there are at least 3 diferences
* rhels <=6.3 * rhels 6.4 * rhels >= 6.5rhels 6.3 is the one that is mainly documented in the xCAT wiki, and if you look through the DRBD/corosync documentation, you will see 2 appendices; A is for 6.4, and B is for 6.5 and above
Before 6.5 the HA software on redhat was technology review, and therefore they made some changed in between. I am hoping that it doesn't get changed after this.
regards, Arif Ali Senior HPC Technical Architect OCF plc Tel: +44 (0)114 257 2200 Mob: +44 (0)7970 148 122 Fax: +44 (0)114 257 0022 Web: www.ocf.co.uk <http://www.ocf.co.uk> Blog: blog.ocf.co.uk <http://blog.ocf.co.uk> Twitter: @ocfplc <http://twitter.com/#%21/ocfplc>OCF plc is a company registered in England and Wales. Registered number 4132533, VAT number GB 780 6803 14. Registered office address: OCF plc, 5 Rotunda Business Centre, Thorncliffe Park, Chapeltown, Sheffield, S35 2PG.
This message is private and confidential. If you have received this message in error, please notify us and remove it from your system.
On 05/06/14 14:50, Lissa Valletta wrote:
No we do not allow multiple active Management nodes.Using servicenodes, we have the xcat daemon (xcatd) running on each service node. Your xcat commands will be run from the servicenodes. You run the commands from the management node and based on your configuration, they get sent by xcatd to the correct servicenode. The Management Node must be active, it has the xCAT database.Whether you want a backup Management Node is based on your needs of your cluster. Can you tolerate some time where you cannot run xCAT? It all depends on how you are using xCAT. I am not sure how many xCAT users invest in a backup management node. I don't expect a MN to fail very often and if it does can you tolerate the recovery time. The important thing is to have the data backed up so when it fails you can get it up and running quickly.Lissa K. Valletta 8-3/B10 Poughkeepsie, NY 12601 (tie 293) 433-3102Inactive hide details for Jaskaran Singh ---06/05/2014 08:45:22 AM---I did read through that and i see you can do HA node optioJaskaran Singh ---06/05/2014 08:45:22 AM---I did read through that and i see you can do HA node option. The document talks about for high availFrom: Jaskaran Singh <[email protected]> To: xCAT Users Mailing list <[email protected]>, Date: 06/05/2014 08:45 AM Subject: Re: [xcat-user] Multiple management node ------------------------------------------------------------------------ I did read through that and i see you can do HA node option.The document talks about for high availablity, that would mean that the second node would sit idle.I am wondering if xcat allows multiple management nodes to be used who can submit commands to the nodes at the same time. I believe multiple management nodes will create race conditions if issuing commands on the nodes at the same time.In the service node concept, do you put the scheduler on each service node for it to be able to submit jobs to the resources.Does xcat recommend having more than 2 management nodes because i dont see it in any architecture diagram?ThanksOn Wed, Jun 4, 2014 at 9:30 PM, Guang Cheng Li <[email protected]_ <mailto:[email protected]>> wrote:HI, The doc _http://sourceforge.net/apps/mediawiki/xcat/index.php?title=Highly_Available_Management_Node_ describes the different HA MN options for xCAT, you could take a look and if you have further question, please let me know. ------------------------------------------------------------------------- Li,Guang Cheng (李光成) IBM China System Technology Laboratory Email: [email protected]_ <mailto:[email protected]> Address: Building 28, ZhongGuanCun Software Park, No.8, Dong Bei Wang West Road, Haidian District Beijing 100193, PRC 北京市海淀区东北旺西路8号中关村软件园28号楼 邮编: 100193 Inactive hide details for Jaskaran Singh ---2014/06/05 04:22:10---I know xcat has the concept for HA of Management and service Jaskaran Singh ---2014/06/05 04:22:10---I know xcat has the concept for HA of Management and service nodes. Has anyone deployed multiple man From: Jaskaran Singh <[email protected]_ <mailto:[email protected]>> To: xCAT Users Mailing list <[email protected]_ <mailto:[email protected]>>, Date: 2014/06/05 04:22 Subject: [xcat-user] Multiple management node ------------------------------------------------------------------------ I know xcat has the concept for HA of Management and service nodes. Has anyone deployed multiple management nodes. I dont see any advantage in having multiple management nodes but i need to research if it is possible, is it recommended, is anyone deploying it. Thanks Jaskaran------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today!_ __http://p.sf.net/sfu/NeoTech________________________________________________ xCAT-user mailing list_ [email protected]_ <mailto:[email protected]>_ __https://lists.sourceforge.net/lists/listinfo/xcat-user_ ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today!_ __http://p.sf.net/sfu/NeoTech_ _______________________________________________ xCAT-user mailing list_ [email protected]_ <mailto:[email protected]>_ __https://lists.sourceforge.net/lists/listinfo/xcat-user_ ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book"Graph Databases" is the definitive new guide to graph databases and theirapplications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/NeoTech_______________________________________________ xCAT-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xcat-user
<<attachment: aali.vcf>>
------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/NeoTech
_______________________________________________ xCAT-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xcat-user
