Tracing Disk Activity In The Linux Kernel Using LTTng

This video walks you through the steps for collecting the relevant set of LTTng events to analyze disk activity on Linux. I will use Tracecompass Input/Output analysis to derive physical disk throughput and disk device queues that will be valuable information in troubleshooting different disk issues.
Music from Uppbeat (free for Creators!):
uppbeat.io/t/s...
License code: F6NB9WE29LWYNPLS

[@amizednix]

Awesome work, please do not stop sharing!

[@ttt1971]

🔥🔥🔥🔥

[@dabbopabblo]

Why is this valuable to know? After making the various observations you made throughout this, what would one do with that acquired information?

[@deeplinux2248]

There are many different directions this could go. You could be interested in this from a purely performance characterization standpoint.
Alternatively, you could be troubleshooting application slowness due to file access which may turn out to be because the file system cache is not performing as expected and all access is directly ending up on the disk. This would be apparent from these analysis.
Or you could realize that there is synchronized disk contention between several applications. Note that such contention is not clearly observable from the file system api consisting of reads and writes because there is a file system cache that skewes, filters, and aggregates event sent to the disk layer.
I hope this provides some clarification.

[@jindiggs]

@@deeplinux2248 I agree with you and I disagree completely with Pabblo.
This is a material usage understanding of filesystems and disk management. There are no concepts here, but rather analysis of something that is used in most servers. What's more crucial is that this video helps one come up with more ideas on how to save time, how to program in faster environments and how to troubleshoot.
I don't want to point any accusations, but those that do not appreciate analysis, are usually superficial. Please keep more content coming. We need this !