More stable kernels with partial Dirty Frag fixes

[lwn@fedi.lwn.net]

More stable kernels with partial Dirty Frag fixes

https://lwn.net/Articles/1071483/ #LWN #Linux #kernel

[aho@ioc.exchange]

@lwn a bit unrelated, the #postgresql performance degradation in the #Linux #kernel 7, has it been fixed (either postgresql or kernel side)?

[tomasv@fosstodon.org]

@aho @lwn I don't think so, or at least I'm not aware of a fix. But I also think the workload showing the regression is extremely artificial, and you'll not see the regression for anything remotely realistic.

[aho@ioc.exchange]

@tomasv @lwn that's good to hear, as had to upgrade kernel and was a bit worried over a headline I saw some time ago when I didn't really planed on upgrading kernel...

[lwn@fedi.lwn.net]

@aho @tomasv This is the article, "The 7.0 scheduler regression that wasn't".

The tl;dr was that the "regression" happened on a system running PostgreSQL with huge pages disabled. Once the reporter enabled them, the problem disappeared.

The 7.0 scheduler regression that wasn't

One of the more significant changes in the 7.0 kernel release is to use the lazy-preemption mod [...]

LWN.net (lwn.net)

[aho@ioc.exchange]

@lwn @tomasv thanks