tag:blogger.com,1999:blog-9238405.post3482457913070659697..comments2024-03-18T02:04:50.380-07:00Comments on Agile Testing: Quick note on installing and configuring GangliaGrig Gheorghiuhttp://www.blogger.com/profile/17863511617654196370noreply@blogger.comBlogger6125tag:blogger.com,1999:blog-9238405.post-8954850182602978272012-03-15T09:40:27.737-07:002012-03-15T09:40:27.737-07:00Could you say which documentation you used to lear...Could you say which documentation you used to learn how to set up your Ganglia configuration? The old IBM wiki, the ganglia.info wiki, or the stuff on github? Did you have to do any troubleshooting, or did it all "just work" out of the box?<br />thanks0.3E9m/shttps://www.blogger.com/profile/14168254673452136370noreply@blogger.comtag:blogger.com,1999:blog-9238405.post-53869286693492843112010-09-22T07:39:40.776-07:002010-09-22T07:39:40.776-07:00Mark - thanks for the tip, I'll definitely try...Mark - thanks for the tip, I'll definitely try that.Grig Gheorghiuhttps://www.blogger.com/profile/17863511617654196370noreply@blogger.comtag:blogger.com,1999:blog-9238405.post-14488955394995268372010-09-22T07:09:32.051-07:002010-09-22T07:09:32.051-07:00great...
Since you're obviously interested ...great... <br /><br />Since you're obviously interested in disk stats and as an experiment you may find useful, log some data locally and then generate a plot file (playback the data with -P -f) and use colplot from collectl-utils to plot the data for your individual disks OR you could always load it into a spreadsheet but depending on thee number of disks you could end up with a lot of columns.<br /><br />If you have a lot of data you can't do this with rrd because of the normalization effect I mentioned earlier, but gnuplot (which colplot) dutifully plots every bit of data with not modification.<br /><br />-markMark Segerhttps://www.blogger.com/profile/05813976850043130168noreply@blogger.comtag:blogger.com,1999:blog-9238405.post-26765193727451415062010-09-22T07:03:11.102-07:002010-09-22T07:03:11.102-07:00Hi Mark -- yes I did post a tweet on collectl. And...Hi Mark -- yes I did post a tweet on collectl. And my intention is to use collectl in conjunction with ganglia, for disk stats as a start. So expect another blog post on that soon ;-)<br /><br />GrigGrig Gheorghiuhttps://www.blogger.com/profile/17863511617654196370noreply@blogger.comtag:blogger.com,1999:blog-9238405.post-28366602299089993272010-09-22T04:36:08.179-07:002010-09-22T04:36:08.179-07:00hey greg - I'm guessing you're the one you...hey greg - I'm guessing you're the one you posted a tweet about collectl the other day (can't hide from google ;)). Anyhow are you aware that you can use collectl to feed data to ganglia? Collect the details on each system and pass a summary to ganglia.<br /><br />Ganglia is pretty with 3 or 4 exceptions:<br />- if you lose the network, you lose the data when you may most need it<br />- the data cannot be gathered frequently enough to deal with subtle problems (I take 5 to 10 second samples)<br />- if you try plotting the data, rrd lies to you and normalizes the data under the covers. I want to see it all every time!<br /><br />-markMark Segerhttps://www.blogger.com/profile/05813976850043130168noreply@blogger.comtag:blogger.com,1999:blog-9238405.post-72817866342971959962010-09-22T04:35:01.079-07:002010-09-22T04:35:01.079-07:00hey greg - I'm guessing you're the one you...hey greg - I'm guessing you're the one you posted a tweet about collectl the other day (can't hide from google ;)). Anyhow are you aware that you can use collectl to feed data to ganglia? Collect the details on each system and pass a summary to ganglia.<br /><br />Ganglia is pretty with 3 or 4 exceptions:<br />- if you lose the network, you lose the data when you may most need it<br />- the data cannot be gathered frequently enough to deal with subtle problems (I take 5 to 10 second samples)<br />- if you try plotting the data, rrd lies to you and normalizes the data under the covers. I want to see it all every time!<br /><br />-markMark Segerhttps://www.blogger.com/profile/05813976850043130168noreply@blogger.com