All-in-one performance analysis with standard Linux tools
In One Fell Swoop
All three listings can be compiled in one go with the commands from Listing 5. The second call creates an executable greggalizer
that runs independently of any shared libraries. So it can be easily copied to related architectures to do its job there without any problems.
Listing 5
Compiling the Listings
$ go mod init greggalizer $ go build greggalizer.go runit.go pane-splitter.go
When this is done, the call to greggalizer
at the command line, as shown in Figure 1, hijacks the terminal, splits it into vertically stacked boxes, and prints the executed commands with parameters and the text output in them. The performance analysis can begin.
Outlook
One way of potentially improving the script presented here would be to parallelize running of the commands, so that the user does not have to wait a few seconds in front of a blank screen until it fills up with results. Instead, the program could first draw the UI, launch the external processes simultaneously via Go routines and continuously keep filling the text widgets with new output via channels, as soon as it becomes available. The result would be a tool like top
that regularly refreshes its widgets.
Infos
- Gregg, Brendan. BPF Performance Tools. Addison-Wesley, 2019
- "Programming Snapshot – bpftrace" by Mike Schilli, Linux Magazine, issue 230, January 2020, pg.56
- Listings for this article: ftp://ftp.linux-magazine.com/pub/listings/linux-magazine.com/235
« Previous 1 2
Buy this article as PDF
(incl. VAT)