Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Mar 18 2020 01:59
    danoagithub starred bklockwood/PSPerf
  • Feb 23 2018 00:03
    posh-cja starred bklockwood/PSPerf
  • May 08 2017 10:04
    jimaek commented #23
  • Jan 31 2017 04:54
  • Mar 24 2016 15:42
  • Mar 12 2016 06:36
    bklockwood opened #55
  • Feb 27 2016 07:08

    bklockwood on master

    Update README.md (compare)

  • Feb 27 2016 07:07

    bklockwood on master

    Update README.md (compare)

  • Feb 14 2016 00:11
    jcislowski starred bklockwood/PSPerf
  • Nov 12 2015 21:40
    bklockwood commented #34
  • Nov 07 2015 15:08

    bklockwood on master

    working to find memory leak (no… Merge pull request #54 from bkl… (compare)

  • Nov 07 2015 15:08
    bklockwood closed #54
  • Nov 07 2015 15:07
    bklockwood opened #54
  • Nov 07 2015 15:07

    bklockwood on bkl-dev

    working to find memory leak (no… (compare)

  • Nov 06 2015 11:16
    bklockwood commented #34
  • Nov 06 2015 11:14

    bklockwood on master

    new psperf.html is cleaner, add… Merge pull request #53 from bkl… (compare)

  • Nov 06 2015 11:14
    bklockwood closed #53
  • Nov 06 2015 11:14
    bklockwood opened #53
  • Nov 06 2015 11:13

    bklockwood on bkl-dev

    new psperf.html is cleaner, add… (compare)

  • Nov 06 2015 07:20

    bklockwood on master

    minor changes to output-page Merge pull request #52 from bkl… (compare)

Oliver Lipkau
@lipkau
in addition: what do you plan on measuring in the events?
Oliver Lipkau
@lipkau
Oliver Lipkau
@lipkau
ignore the template.
I worked on the graphics
http://instr.de/PSPerf/v1/#
http://instr.de/PSPerf/v1/data/test.json
Bryan Lockwood
@bklockwood
I know this is not the common wisdom, but I feel that %processor time is a nearly useless number. The processor can be at 99% and still be servicing all threads as rapidly as if it were at 30%. CPU queue measures how many threads are waiting for service and is thus a true measure of whether the CPU is a bottleneck.
All of the metrics I choose are based on this concept. See the concept page in the project wiki.
Your http://instr.de/PSPerf/v1/# is interesting to me but it is very large. My focus is on a sysadmin (or group) who have multiple servers to monitor, and want a "big board" that can give them an at-a-glance understanding of which of those systems are healthy, and which ones are wallowing.
Bryan Lockwood
@bklockwood
So it should be able to show a bunch of systems results on one page. If possible I want to make each server line smaller - as long as there's enough clarity to see which monitored system needs attention.
Events. I'm not sure yet but I think I will probably collect the number of new error/warning events in the System and Application logs, and graph that. I have a hazy methodology in mind for this, not nailed down yet. It will probably be the next thing I work on, after I kill a bug I seem to have created in last night's code additions.
Oliver Lipkau
@lipkau
With your ps1 generated HTML you can -join "," the array with the values into the <div class
Bryan Lockwood
@bklockwood
that's what it does? example: $Output += "<td><span class=""cpu"">$($StorageHash.$PC.CpuQueue -join(","))</span></td>rn"
(line 439)
Or do you mean I should be using div rather than td?
Oliver Lipkau
@lipkau
That was a repeated message
Don't know why gitter sent it
Bryan Lockwood
@bklockwood
ah, ok
Oliver Lipkau
@lipkau
I am also not happy with my line height
But I also don't like the idea adding a new row for each disk
Bryan Lockwood
@bklockwood
yeah. My advice to users will probably be "do not monitor more than 2 disks per server"
Oliver Lipkau
@lipkau
Creating the dynamic row addition according to max disks for a computer is not hard, but will explode in the horizontal axis
And restricting to 2 disks is a lazy fix :-P
We will figure something out
Tomorrow I will have a few hours to code. Looking forward to it
Bryan Lockwood
@bklockwood
well, yes ... and no. I feel that sometimes limits help focus goals. And here the main goal is not "make a lot of pretty graphic", it is "show the minimum amount of data that indicates the server's overall health"
Oliver Lipkau
@lipkau
It's been a while
Bryan Lockwood
@bklockwood
So, as an experienced server admin, I feel that there are few cases where you need more than two disks to show how healthy a server is (typically those with pagefiles on them). I plan some guidance opn this in final product.
Oliver Lipkau
@lipkau
You could as the server twice with different disks :-P
Bryan Lockwood
@bklockwood
it's a possibility, yes
or, maybe just collect queue lengths for all disks, and only show the one with the worst looking graph
Oliver Lipkau
@lipkau
Does get-counter require wsman on the target?
Bryan Lockwood
@bklockwood
no. in fact I've researched this a bit and no one seems clear on exactly how get-counter works. I know it does not use wsman. see https://github.com/bklockwood/PSPerf/wiki/Retrieving-Perfmon-Counters-Remotely
Oliver Lipkau
@lipkau
I am not used to coders who document stuff that of in dev :-P
Bryan Lockwood
@bklockwood
I will be looking for more info on this. I could swear that I have or used to have a book that goes into more depth on it; perfmon is old
I am big on docs, always have been. I doc as I go because I know that if I don't, there won;t be docs.
also I have never had great memory
Oliver Lipkau
@lipkau
I like the 'the doc is in the code' approach
Bryan Lockwood
@bklockwood
Coders always do. Lots of sysadmins hate it.
Because doc-in-code is usually kinda crappy. Forcing a document that actually talks to the user is important, imo.
Oliver Lipkau
@lipkau
I like ps for the auto help generation for functions
I wrote a PsADManagement module for which I invested a lot of time for the function helpers
Bryan Lockwood
@bklockwood
yes, yes, yes. it does seem to be encouraging better doc-in-code.
yours, btw, is exemplary from what I have seen so far.
Oliver Lipkau
@lipkau
Did you see that module?
Bryan Lockwood
@bklockwood
saw it, but have not explored it
Oliver Lipkau
@lipkau
I started refactoring your data gathering for JSON output.
I added a functionality which always delivers $maxEntries (default 50) values. So the LineGraphs don't flatline while the is no history.
and it adds value 0 when computer is offline
https://github.com/lipkau/PSPerf
I didn't make any of the changes I wanted to the HTML. I didn't touch the HTML and it outputs like this to the generated JSON:
http://instr.de/PSPerf/v2/
(not much going on in the graphs, as it is real data from my server which is not overloaded)
Bryan Lockwood
@bklockwood
Hi Oliver. I looked over your code a bit last night. Very impressive.
(psperf I mean). I will Very likely use your ideas for storing as json, and refactoring the data collection a bit. But not quite yet, I still want to get a couple of other things done in the codebase I have.
Have you tested your existing code for gathering data from other systems?
Oliver Lipkau
@lipkau
I have not tested it on remote computers.
I don't have a setup which allows me to do so. I must first setup a domain in hyper-v