[OpenAFS] ramdisk Problems under 2.6 kernel

ted creedon tcreedon@easystreet.com
Wed, 6 Jul 2005 08:18:14 -0700


This is a multi-part message in MIME format.

------=_NextPart_000_02C1_01C58203.4203B860
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit

It might be illustrative to use the MIT stress tests described at the 2005
Conference to benchmark various alternatives.
 
The presenter indicated that the test sources may be available in a few
months.
 
tedc

  _____  

From: openafs-info-admin@openafs.org [mailto:openafs-info-admin@openafs.org]
On Behalf Of ted creedon
Sent: Wednesday, July 06, 2005 7:47 AM
Cc: openafs-info@openafs.org
Subject: RE: [OpenAFS] ramdisk Problems under 2.6 kernel


Precisely what I do for a test system. Loopbacks work fine. 
 
Any serious performance hits with loopback devices on /dev/shm?
 
tedc

  _____  

From: Sven Oehme [mailto:oehmes@de.ibm.com] 
Sent: Wednesday, July 06, 2005 12:52 AM
To: ted creedon
Cc: openafs-info@openafs.org
Subject: RE: [OpenAFS] ramdisk Problems under 2.6 kernel



because you don't reach high enough throughput with memcache. 
with memcache i am limited to around 25 mb/sec. with disk cache i reach 30
mb/sec which is the local drive. 
with a formatted ramdisk as diskcache i get 50 - 60 mb/sec 
so lets see if we can get a fix for /dev/shm ... because ramdisk is anyway
deprecated. 
btw. the workaround is to create a file in /dev/shm with dd, format it ,
mount it as loopdevice :-) 

Sven 




"ted creedon" <tcreedon@easystreet.com> 
Sent by: openafs-info-admin@openafs.org 


05/07/2005 21:59 


To

cc
<openafs-info@openafs.org> 

Subject
RE: [OpenAFS] ramdisk Problems under 2.6 kernel	

		




Why is that? 
  
tedc 


  _____  

From: openafs-info-admin@openafs.org [mailto:openafs-info-admin@openafs.org]
On Behalf Of Sven Oehme
Sent: Tuesday, July 05, 2005 10:08 AM
To: Ryan Underwood
Cc: openafs-info@openafs.org
Subject: Re: [OpenAFS] ramdisk Problems under 2.6 kernel


it's simple , it's to slow .. 

Sven



Ryan Underwood <nemesis-lists@icequake.net> 
Sent by: openafs-info-admin@openafs.org 


05/07/2005 17:46 




To
openafs-info@openafs.org 

cc

Subject
Re: [OpenAFS] ramdisk Problems under 2.6 kernel	


		






On Mon, Jul 04, 2005 at 11:38:14PM +0200, Sven Oehme wrote:
>    Hi ,
> 
>    we have identified some Ramdisk Problems under 2.6 kernels compiled on
>    i386
>    has anybody ever tried using tmpfs for the cache disk ?

What about using -memcache instead?

-- 
Ryan Underwood, <nemesis@icequake.net>
_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info



------=_NextPart_000_02C1_01C58203.4203B860
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dus-ascii">
<META content=3D"MSHTML 6.00.2900.2668" name=3DGENERATOR></HEAD>
<BODY>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D093441515-06072005>It might be illustrative to use the MIT =
stress tests=20
described at the 2005 Conference to benchmark various=20
alternatives.</SPAN></FONT></DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D093441515-06072005></SPAN></FONT>&nbsp;</DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D093441515-06072005>The presenter indicated that the test sources =
may be=20
available in a few months.</SPAN></FONT></DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D093441515-06072005></SPAN></FONT>&nbsp;</DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D093441515-06072005>tedc</SPAN></FONT></DIV><BR>
<DIV class=3DOutlookMessageHeader lang=3Den-us dir=3Dltr align=3Dleft>
<HR tabIndex=3D-1>
<FONT face=3DTahoma size=3D2><B>From:</B> openafs-info-admin@openafs.org =

[mailto:openafs-info-admin@openafs.org] <B>On Behalf Of </B>ted=20
creedon<BR><B>Sent:</B> Wednesday, July 06, 2005 7:47 AM<BR><B>Cc:</B>=20
openafs-info@openafs.org<BR><B>Subject:</B> RE: [OpenAFS] ramdisk =
Problems under=20
2.6 kernel<BR></FONT><BR></DIV>
<DIV></DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D011214514-06072005>Precisely what I do for a test system. =
Loopbacks work=20
fine. </SPAN></FONT></DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D011214514-06072005></SPAN></FONT>&nbsp;</DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D011214514-06072005>Any serious performance hits with loopback =
devices on=20
/dev/shm?</SPAN></FONT></DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D011214514-06072005></SPAN></FONT>&nbsp;</DIV>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D011214514-06072005>tedc</SPAN></FONT></DIV><BR>
<DIV class=3DOutlookMessageHeader lang=3Den-us dir=3Dltr align=3Dleft>
<HR tabIndex=3D-1>
<FONT face=3DTahoma size=3D2><B>From:</B> Sven Oehme =
[mailto:oehmes@de.ibm.com]=20
<BR><B>Sent:</B> Wednesday, July 06, 2005 12:52 AM<BR><B>To:</B> ted=20
creedon<BR><B>Cc:</B> openafs-info@openafs.org<BR><B>Subject:</B> RE: =
[OpenAFS]=20
ramdisk Problems under 2.6 kernel<BR></FONT><BR></DIV>
<DIV></DIV><BR><FONT face=3Dsans-serif size=3D2>because you don't reach =
high enough=20
throughput with memcache. </FONT><BR><FONT face=3Dsans-serif =
size=3D2>with memcache=20
i am limited to around 25 mb/sec. with disk cache i reach 30 mb/sec =
which is the=20
local drive.</FONT> <BR><FONT face=3Dsans-serif size=3D2>with a =
formatted ramdisk as=20
diskcache i get 50 - 60 mb/sec</FONT> <BR><FONT face=3Dsans-serif =
size=3D2>so lets=20
see if we can get a fix for /dev/shm ... because ramdisk is anyway =
deprecated.=20
</FONT><BR><FONT face=3Dsans-serif size=3D2>btw. the workaround is to =
create a file=20
in /dev/shm with dd, format it , mount it as loopdevice :-)</FONT> =
<BR><BR><FONT=20
face=3Dsans-serif size=3D2>Sven</FONT> <BR><BR><BR><BR>
<TABLE width=3D"100%">
  <TBODY>
  <TR vAlign=3Dtop>
    <TD width=3D"40%"><FONT face=3Dsans-serif size=3D1><B>"ted creedon"=20
      &lt;tcreedon@easystreet.com&gt;</B> </FONT><BR><FONT =
face=3Dsans-serif=20
      size=3D1>Sent by: openafs-info-admin@openafs.org</FONT>=20
      <P><FONT face=3Dsans-serif size=3D1>05/07/2005 21:59</FONT> </P>
    <TD width=3D"59%">
      <TABLE width=3D"100%">
        <TBODY>
        <TR>
          <TD>
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>To</FONT></DIV>
          <TD vAlign=3Dtop>
        <TR>
          <TD>
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>cc</FONT></DIV>
          <TD vAlign=3Dtop><FONT face=3Dsans-serif=20
            size=3D1>&lt;openafs-info@openafs.org&gt;</FONT>=20
        <TR>
          <TD>
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>Subject</FONT></DIV>
          <TD vAlign=3Dtop><FONT face=3Dsans-serif size=3D1>RE: =
[OpenAFS] ramdisk=20
            Problems under 2.6 =
kernel</FONT></TD></TR></TBODY></TABLE><BR>
      <TABLE>
        <TBODY>
        <TR vAlign=3Dtop>
          <TD>
          =
<TD></TD></TR></TBODY></TABLE><BR></TD></TR></TBODY></TABLE><BR><BR><BR><=
FONT=20
face=3DArial color=3Dblue size=3D2>Why is that?</FONT> <BR><FONT =
size=3D3>&nbsp;</FONT>=20
<BR><FONT face=3DArial color=3Dblue size=3D2>tedc</FONT> <BR><BR>
<HR>
<FONT face=3DTahoma size=3D2><B>From:</B> openafs-info-admin@openafs.org =

[mailto:openafs-info-admin@openafs.org] <B>On Behalf Of </B>Sven=20
Oehme<B><BR>Sent:</B> Tuesday, July 05, 2005 10:08 AM<B><BR>To:</B> Ryan =

Underwood<B><BR>Cc:</B> openafs-info@openafs.org<B><BR>Subject:</B> Re:=20
[OpenAFS] ramdisk Problems under 2.6 kernel</FONT><FONT=20
size=3D3><BR></FONT><BR><FONT face=3Dsans-serif size=3D2><BR>it's simple =
, it's to=20
slow ..</FONT><FONT size=3D3> <BR></FONT><FONT face=3Dsans-serif=20
size=3D2><BR>Sven</FONT><FONT size=3D3><BR><BR><BR></FONT>
<TABLE width=3D"100%">
  <TBODY>
  <TR vAlign=3Dtop>
    <TD width=3D"44%"><FONT face=3Dsans-serif size=3D1><B>Ryan Underwood =

      &lt;nemesis-lists@icequake.net&gt;</B> <BR>Sent by:=20
      openafs-info-admin@openafs.org</FONT><FONT size=3D3> </FONT>
      <P><FONT face=3Dsans-serif size=3D1>05/07/2005 17:46</FONT><FONT =
size=3D3>=20
      </FONT></P>
    <TD width=3D"55%"><BR>
      <TABLE width=3D"100%">
        <TBODY>
        <TR>
          <TD width=3D"18%">
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>To</FONT></DIV>
          <TD vAlign=3Dtop width=3D"81%"><FONT face=3Dsans-serif=20
            size=3D1>openafs-info@openafs.org</FONT><FONT size=3D3> =
</FONT>
        <TR>
          <TD>
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>cc</FONT></DIV>
          <TD vAlign=3Dtop>
        <TR>
          <TD>
            <DIV align=3Dright><FONT face=3Dsans-serif =
size=3D1>Subject</FONT></DIV>
          <TD vAlign=3Dtop><FONT face=3Dsans-serif size=3D1>Re: =
[OpenAFS] ramdisk=20
            Problems under 2.6 =
kernel</FONT></TD></TR></TBODY></TABLE><BR><BR>
      <TABLE width=3D"100%">
        <TBODY>
        <TR vAlign=3Dtop>
          <TD width=3D"49%">
          <TD=20
width=3D"50%"></TD></TR></TBODY></TABLE><BR></TD></TR></TBODY></TABLE><BR=
><FONT=20
size=3D3><BR><BR></FONT><FONT size=3D2><TT><BR><BR>On Mon, Jul 04, 2005 =
at=20
11:38:14PM +0200, Sven Oehme wrote:<BR>&gt; &nbsp; &nbsp;Hi ,<BR>&gt; =
<BR>&gt;=20
&nbsp; &nbsp;we have identified some Ramdisk Problems under 2.6 kernels =
compiled=20
on<BR>&gt; &nbsp; &nbsp;i386<BR>&gt; &nbsp; &nbsp;has anybody ever tried =
using=20
tmpfs for the cache disk ?<BR><BR>What about using -memcache =
instead?<BR><BR>--=20
<BR>Ryan Underwood,=20
&lt;nemesis@icequake.net&gt;<BR>_________________________________________=
______<BR>OpenAFS-info=20
mailing=20
list<BR>OpenAFS-info@openafs.org<BR>https://lists.openafs.org/mailman/lis=
tinfo/openafs-info</TT></FONT><FONT=20
size=3D3><BR></FONT><BR></BODY></HTML>

------=_NextPart_000_02C1_01C58203.4203B860--