Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 23 Apr 2010 03:33:03 +0300
From:      Andriy Gapon <avg@icyb.net.ua>
To:        jhell <jhell@dataix.net>
Cc:        freebsd-fs@freebsd.org
Subject:   Re: some arc_reclaim_needed stats
Message-ID:  <4BD0EABF.6030100@icyb.net.ua>
In-Reply-To: <4BD0E8A9.4040403@dataix.net>
References:  <4BC86CF3.7060708@icyb.net.ua> <4BCEF7E4.6080606@dataix.net> <4BD0BC44.1040204@icyb.net.ua> <4BD0E8A9.4040403@dataix.net>

next in thread | previous in thread | raw e-mail | index | archive | help
on 23/04/2010 03:24 jhell said the following:
> On 04/22/2010 17:14, Andriy Gapon wrote:
>> on 21/04/2010 16:04 jhell said the following:
>>> If these are not tunable values and serve a good purpose as a stat to be
>>> added at some point, would they not be better in kstat.zfs.misc? rather
>>> than vfs.zfs so they can be collected with the rest of the stats via
>>> arc_summary.pl.
>> Yes, I agree.
>> But right now no one seems to think that there is anything useful about these
>> stats :-)
>>
>>> Without seeing the thread "kstat.zfs.misc.arcstats.hash_collisions" I
>>> had updated the arc_summary.pl script yesterday to include these under a
>>> "ARC Misc" section in the output. Just a FYI in case it may be useful to
>>> you or others.
>> BTW, it would be nice if hash table size (number of buckets) would be reported
>> too.  But that's a kernel change, not arc_summary.
>>
> 
> Something like what is in r59 and in the download section ?
> 
> ARC Hash Breakdown:
> 	Elements Max:				129057
> 	Elements Current:		99.24%	128070
> 	Collisions:				251021
> 	Chain Max:				21
> 	Chains:					16326

No, I mean number of the buckets (not elements) in the hash table.
http://en.wikipedia.org/wiki/Hash_table
:-)

-- 
Andriy Gapon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BD0EABF.6030100>