[Top] [All Lists]

Re: Hung in D state during fclose

To: "Cheung, Norman" <Norman.Cheung@xxxxxxxxxxxxxx>
Subject: Re: Hung in D state during fclose
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Wed, 13 Feb 2013 09:21:50 +1100
Cc: "linux-xfs@xxxxxxxxxxx" <linux-xfs@xxxxxxxxxxx>
Delivered-to: linux-xfs@xxxxxxxxxxx
In-reply-to: <3542214BE3A3EF419F236DFE0F878BC905190D@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
References: <loom.20130212T071115-446@xxxxxxxxxxxxxx> <20130212065545.GC10731@dastard> <3542214BE3A3EF419F236DFE0F878BC90512DC@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <20130212102014.GA26694@dastard> <3542214BE3A3EF419F236DFE0F878BC90517D2@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <20130212202246.GB26694@dastard> <3542214BE3A3EF419F236DFE0F878BC905190D@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Tue, Feb 12, 2013 at 09:06:03PM +0000, Cheung, Norman wrote:
> [NLC] Only 2 threads in D state.  I  need to wait for the next
> hang to take another stack trace.  Is there any workaround I can
> reduce the frequencies of this hang.  What about reducing the
> xfssyncd_centisecs? Or other knobs?

You need to isolate the cause of the hang before changing anything.
Twiddling random knobs won't improve the situation. Understand the
problem first, then take appropriate action.

> [NLC]  2 RAID controllers in the system, it hung from disks on
> both of them.  One is SuperMicro 2208 and the other is LSI
> 9265-8I.  I think both use the same chipset.

Does the filesystem that hangs always belong to the same controller?


Dave Chinner

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