On 05/11/2011 10:47 AM, Burnash, James wrote: > Hi Joe. > > Your remarks are always useful and informative - thanks. Thanks for the compliment! > As for our slow network throughput - what I didn't put in our > configuration is that our servers are on 10GBe, but all of our > clients are on 1Gbe because our core network can't (yet) handle the > load of 10GBe clients - just to fill in that data point. Got it, that makes sense. We see typically 30-80% of wire speed in connections (depending upon contention and other things). Your numbers fall into that range. > I find your remark about the stability of 3.1.3 reassuring - > considering the painful struggle to get there from the 3.0.4 > versions, and the stability issues that I noted in the list over the 3.0.5 has been remarkably stable at customer sites ... no complaints. 3.1.x has been a struggle until 3.1.2 and 3.1.3. Ran head first into some 3.1.4 issues that I still cannot tell if they were migration issues or real bugs. 3.2.0 was a test effort that did not succeed internally, so we backed off for the moment. > course of my migration. Your problems with 3.1.4 and 3.2 are enough > reason to not do anymore upgrades to the production systems yet. > We always recommend staging upgrades on test machines if possible. Sometimes you get bitten by some nasty bits you were not expecting (not with Gluster per se, but with an odd interaction). > As a rule of thumb, I never implement X.0 releases into production > anyhow - even from Redhat ... I have the arrows still sticking out of > my back from doing so in the past :-) Heh ... I am pretty happy so far with Centos/RHEL 5.6. We haven't tested the 6.0 much yet. Will do that soon. -- Joseph Landman, Ph.D Founder and CEO Scalable Informatics Inc. email: landman at scalableinformatics.com web : http://scalableinformatics.com http://scalableinformatics.com/sicluster phone: +1 734 786 8423 x121 fax : +1 866 888 3112 cell : +1 734 612 4615
