xfs-masters
[Top] [All Lists]

[Bug 16348] kswapd continuously active when doing IO

To: xfs-masters@xxxxxxxxxxx
Subject: [Bug 16348] kswapd continuously active when doing IO
From: bugzilla-daemon@xxxxxxxxxxxxxxxxxxx
Date: Tue, 13 Jul 2010 08:15:54 GMT
Auto-submitted: auto-generated
In-reply-to: <bug-16348-470@xxxxxxxxxxxxxxxxxxxxxxxxx/>
References: <bug-16348-470@xxxxxxxxxxxxxxxxxxxxxxxxx/>
https://bugzilla.kernel.org/show_bug.cgi?id=16348





--- Comment #11 from Dave Chinner <david@xxxxxxxxxxxxx>  2010-07-13 08:15:45 ---
Ok, so there's about 4m clean cached inodes and dentries at the start, and 2.5m
cached dentries and 3.6m cached inodes at the end. reclaim doesn't really seem
to be making much progress here - inodes can't be reclaimed until dentries are
freed, and that isn't happening very fast, either.

I can reproduce kswapd running for a few seconds while it reclaims a couple of
million inodes and dentries, but I can't keep it running for minutes. 

Right now I'll focus on the xfs-perag_get/put frequency and try to work that
out. Can you post the output of xfs_info for each XFS filesystem you have so i
can see how many filesystems and AGs are being traversed during each reclaim
call...

-- 
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

<Prev in Thread] Current Thread [Next in Thread>