Linux clustering is a steaming pile of shit

Jan 04, 2010 02:50

[To longtime readers of my blog, sorry about the more CS-y content that's starting to appear. It's all part of Iron Blogger. I'll try to also make sure to post some more personal stuff ( Read more... )

mit, linux, sipb

Leave a comment

Comments 3

Corrections anonymous January 4 2010, 13:45:20 UTC
Get your facts straight before mouthing off.

Pacemaker is not a fork of Heartbeat:
Heartbeat 3.0 = Heartbeat 2.x - Pacemaker 0.6 - ClusterGlue 1.0 - ResourceAgents 1.0

Corosync is not a fork of openais:
OpenAIS 1.0 = OpenAIS 0.80 - Corosync 1.0

In both cases there was no fork, only the modularization of monolithic codebases.
The code continues to be maintained by its original authors.

There is also no Heartbeat "development".
Linbit is simply releasing whats left after the splits with the intention of fixing any serious bugs that come up.

Reply

Re: Corrections coolerq January 4 2010, 15:23:27 UTC
Sorry, you're right that my post isn't clear; "fork" implies that there is a difference between the new version's developers and the old version's. I've updated the post to just use the word split instead of the word fork.

Heartbeat 3 development definitely seems to have picked up though; I remember 6-12 months ago when the Heartbeat website didn't even work for a period of several months. Now they have a new wiki and are doing regular releases.

--Quentin

Reply


anonymous February 2 2010, 06:17:22 UTC
Hi Quentin!

It's interesting to see "iSCSI-connected RAID" as one of the application on clustering. Do you have any idea about SAS network instead of iSCSI? (How can we make clustering working SAS newtwork?)

--Prashanth (prahanths2012@gmail.com)

Reply


Leave a comment

Up