Re: Increased read IO wait times after Bullseye upgrade

[Vukovics Mihály <vm@informatik.hu>]

Hello Gareth,

the average io wait state is 3% in the last 1d14h. I have checked the IO 
usage with several tools and have not found any processes/threads 
generation too much read/write requests. As you can see on my first 
graph, only the read wait time increased significantly, the write not.

Br,
Mihaly

On 2022. 11. 10. 16:13, Gareth Evans wrote:
> On Thu 10 Nov 2022, at 11:36, Gareth Evans <donotspam@fastmail.fm> wrote:
> [...]
> > I might be barking up the wrong tree ...
> But simpler inquiries first.
>
> I was wondering if MD might be too high-level to cause what does seem more like a "scheduley" issue -
>
> https://www.thomas-krenn.com/de/wikiDE/images/d/d0/Linux-storage-stack-diagram_v4.10.pdf
>
> but, apparently, even relatively "normal" processes can cause high iowait too:
>
> NFS...
> https://www.howtouselinux.com/post/troubleshoot-high-iowait-issue-on-linux-system
> https://www.howtouselinux.com/post/use-linux-nfsiostat-to-troubleshoot-nfs-performance-issue
>
> SSH...
> Use iotop to find io-hogs...
> https://www.howtouselinux.com/post/quick-guide-to-fix-linux-iowait-issue
> https://www.howtouselinux.com/post/check-disk-io-usage-per-process-with-iotop-on-linux
>
> What's a typical wa value (%) from top?
>
> Thanks,
> G
--
--
Köszönettel:
Vukovics Mihály