12-07-2011
To minimize the danger of split brain usually more than one channel for heartbeat is used. The HA-clusters I worked with used 3: user-lan, a seperate cluster-lan and disks. With this configuration I never encountered a split brain situation.
Why would you want to change the ip-addresses during a failover? A simple solution for this would be not to make the ip-addresses cluster resources but configure them locally.
6 More Discussions You Might Find Interesting
1. UNIX for Dummies Questions & Answers
first off, i am new to unix so please bear with me. i was reading somewhere that if your i-nodes get critical that it can slow your network down. what are i-nodes and when do they become a critical number? this is what mine states:
/ (/dev/root ): 777058 blocks 569290 i-nodes... (4 Replies)
Discussion started by: djatwork
4 Replies
2. UNIX for Dummies Questions & Answers
how do you list all the nodes in unix
:confused: (3 Replies)
Discussion started by: kamisi
3 Replies
3. UNIX for Advanced & Expert Users
Does anyone know something about this? I have no idea what it means and how to do it. but if anyone can give me and explanation and also point me to a website, i'd really appreciate it (5 Replies)
Discussion started by: TRUEST
5 Replies
4. UNIX for Advanced & Expert Users
After rcp -rp from remote host, using du -k to verify the file size but total file size have different size. Check on individual file, file size is correct.
How can I confirm on the file size after ftp?
Pls advise.
Thank you. (15 Replies)
Discussion started by: KhawHL
15 Replies
5. High Performance Computing
I am new to cluster commands. But I have tried utilizing: -pe, but I do not know my parallel computing environment. We are running SGE, is there a simpler command to request more nodes?
also: qsub -N auto -M name@email.com -m abe auto.sh
does not email me at all.
Help would be appreciated (0 Replies)
Discussion started by: theawknewbie
0 Replies
6. UNIX for Beginners Questions & Answers
Hi all
I would like to know how to run task2.service after task1.service has finished.
task1.service has a timer (task1.timer), that makes it run every 5 minutes
OnCalendar=*:0/5task2.service is basically a script, that has to work on the files created after task1 has finished.
This is what I... (2 Replies)
Discussion started by: guilliber
2 Replies
LEARN ABOUT CENTOS
votequorum_overview
VOTEQUORUM_OVERVIEW(8) Corosync Cluster Engine Programmer's Manual VOTEQUORUM_OVERVIEW(8)
NAME
votequorum_overview - Votequorum Library Overview
OVERVIEW
The votequorum library is delivered with the corosync project. It is the external interface to the vote-based quorum service. This service
is optionally loaded into all nodes in a corosync cluster to avoid split-brain situations. It does this by having a number of votes
assigned to each system in the cluster and ensuring that only when a majority of the votes are present, cluster operations are allowed to
proceed.
The library provides a mechanism to:
* Query the quorum status
* Get a list of nodes known to the quorum service
* Receive notifications of quorum state changes
* Change the number of votes assigned to a node
* Change the number of expected votes for a cluster to be quorate
* Connect an additional quorum device to allow small clusters to remain quorate during node outages.
BUGS
No known bugs at the time of writing. The authors are from outerspace. Deal with it.
SEE ALSO
corosync-quorumtool(8), votequorum(5), votequorum_initialize(3), votequorum_finalize(3), votequorum_getinfo(3), votequorum_trackstart(3),
votequorum_trackstop(3), votequorum_fd_get(3), votequorum_dispatch(3), votequorum_context_set(3), votequorum_context_get(3), votequo-
rum_setexpected(3), votequorum_setvotes(3)
corosync Man Page 2012-01-12 VOTEQUORUM_OVERVIEW(8)