Thanks a lot. Should I decide to use nfs4, do you think doing what follows could be a good idea?
- I already have 2 active/passive nodes. I could add a resource mounting the storage on them. - I would use nfs4 in order to export the desired data on the outside. I already tried it manually. But my concern is about the reliability of the solution server side and client side in the following cases: A) when I try to swap from master to slave the storage doesn't umount, probably because of some client have already mounted the exported partition. Is there a way to force the swap in a secure way? B) if I stop the nfs services on the server, the clients hang up; this is not desirable. I've bben searching around and someone suggests to mount the remote export using the options bg,intr,hard. I don't appreciate any significant change of (bad) behavior. Server side, what RA should I be using? Where can I find the nfs-kernel-server rpm? If I used the nfsserver RA, should I really share on the storage some folders (such as /var/lib/nfs and other stuff)? Client side, what kind of controls should be done in order to alert that the remote storage is not available, with the following desired options: 1. No hangup; 2. Automatic remount as soon as the nfs server (even if on the backup node) is available again. Thanks for your time G. -----Messaggio originale----- Da: [email protected] [mailto:[email protected]] Per conto di Andrew Beekhof Inviato: venerdì 14 giugno 2013 04:05 A: General Linux-HA mailing list Oggetto: Re: [Linux-HA] Pacemaker & gfs2 on RHEL6.2 x64 On 14/06/2013, at 3:57 AM, Guglielmo Abbruzzese <[email protected]> wrote: > Hi, > I'm working on a issue from a while but it seems to me I've got to a > dead end :( > > Till now, I've been able to setup and configure several cluster > solutions on > RHEL6.2x64 using the RH's Corosync (1.4.1-4) and Pacemaker (1.1.6-3) rpms. > I've started to work on installing a SAN storage, I'd like to use a > shared file system because 8 servers need to store something on it at > the same time. > OCFS2 seems to be not available anymore for my distro. I don't believe OCFS2 was ever available on RHEL. > NFS is not suitable. > So I decided to try and use gfs2, even because of the excellent > "Pacemaker-1.1-Clusters_from_Scratch" install and configuration guide. > > Here are some troubles. > > 1. The user guide refers to Fedora13. It suggests to install > gfs2-utils, cman and gfs2-cluster. In particular, gfs2-cluster is not > provided with RHEL, and the version I've found > (gfs2-cluster-3.1.1-2.fc14.x86_64.rpm) > requires cman >= 3.1.0-1 and openais >= 1.1.2-1; not even RHEL6.4 has > got such versions. What's the best practice in this case? Those packages do not appear in the base distro. You have to pay for the HA channel to get those. Or, follow http://clusterlabs.org/quickstart-redhat.html > > 2. Pacemaker is provided with RHEL and so I thought it was supported > in some way by the RHEL's team (at least about the compatibility with > the other packages); but it isn't actually, and the support has > strongly suggested to swap to Red Hat Cluster Suite. I've been working > a lot on Pacemaker, this is a choice I really wouldn't take... It is slowly becoming less unsupported with every 6.x release. That I need to speak cryptically about such developments is highly annoying. To put it differently, if you run pacemaker on top of cman in 6.4 and find a serious bug, it should be possible push out an update. > > 3. In the worst case, any suggestion about mounting and managing a > shared SAN storage file system with a different solution without > trashing all my job till now for switching to RHCS? > > I'd appreciate a lot if you shared your experience Thank you for your > time Guglielmo > > > > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
