Re: [Pacemaker] clmvd hangs on node1 if node2 is fenced

2010-08-16 Thread Xinwei Hu
2010/8/16 Rainer Lutz : > Xinwei Hu writes: > >> >> This sounds a like a fixed issue for SLE11SP1 indeed. >> referring to: >> http://archives.free.net.ph/message/20100309.075255.baa4b2eb.en.html >> > > Well it is not fixed with SP1, but with some Patch

Re: [Pacemaker] clmvd hangs on node1 if node2 is fenced

2010-08-15 Thread Xinwei Hu
This sounds a like a fixed issue for SLE11SP1 indeed. referring to: http://archives.free.net.ph/message/20100309.075255.baa4b2eb.en.html 2010/8/15 Rainer Lutz : >> Any Idea how to solve it? > > I created a testscenario with the same conditions ( same OS Sles11 SP1 with > HAE > and the CIB from th

Re: [Pacemaker] ra script for clvmd?

2009-04-26 Thread Xinwei Hu
Hi Karl, Here's the RA. ;) 2009/4/27 Karl Katzke : > I'm working with OpenSuSE 11.1 and trying to set up clvmd with pacemaker and > openais. The strange thing is that it *works*, with just pacemaker running, Because the distributed locking in openais is not reliable, clvmd will fail when some

Re: [Pacemaker] i am very confused

2009-02-24 Thread Xinwei Hu
2009/2/20 Glory Smith : > Thanks Lar for reply, >> >> The SBD mechanism provides a way to fence errant nodes. The cluster >> manager itself ensures that it does not activate the resource several >> times. > > my concern is suppose SBD resource is running on say node1 even then node2 > who is member

Re: [Pacemaker] SFEX resource agent

2009-02-24 Thread Xinwei Hu
Here's the attachment... Sorry for push the "sending" button too early. 2009/2/24 Xinwei Hu : > Hi Priyanka, > > 2009/2/22 Priyanka Ranjan : >>> It is close to impossible to protect against malicious applications from >>> other cluster nodes. sfex,

Re: [Pacemaker] SFEX resource agent

2009-02-24 Thread Xinwei Hu
Hi Priyanka, 2009/2/22 Priyanka Ranjan : >> It is close to impossible to protect against malicious applications from >> other cluster nodes. sfex, LVM exclusive activation, and SCSI2/SCSI3 >> reservations likewise can be broken - they must be able to be broken, or >> else the cluster could never o