Re: Snapshot performances

To: Emmanuel Dreyfus <manu%netbsd.org@localhost>, tech-kern%netbsd.org@localhost
Subject: Re: Snapshot performances
From: Taylor R Campbell <campbell+netbsd-tech-kern%mumble.net@localhost>
Date: Mon, 25 Sep 2017 07:54:00 +0000

Emmanuel Dreyfus:
> On a busy mail server running NetBSD 7.1, I can reliabily reproduce
> the following situation when using FFS snapshots: most processes get
> stuck into tstile, load quickly raises to an unreasonable level,
> nothing responds. I have not looked into the internals, but everything 
> behaves likes if all I/O were serialized.

Can you get kernel stack traces from the stuck processes using crash(8)?

Does it unstick, or does it remain stuck forever?  That is, what makes
you conclude `everything behaves like if all I/O were serialized'?

Follow-Ups:
- Re: Snapshot performances
  - From: Emmanuel Dreyfus

References:
- Snapshot performances
  - From: Emmanuel Dreyfus

Prev by Date: Snapshot performances
Next by Date: Re: Snapshot performances
Previous by Thread: Snapshot performances
Next by Thread: Re: Snapshot performances
Indexes:

Home | Main Index | Thread Index | Old Index