[OpenAFS] Client Cache Question

[Timothy Balcer]

--089e013a07d01501f504dfb264d5
Content-Type: text/plain; charset=ISO-8859-1

I have a tiny head scratcher...

Tested this case:

rsync 100 or so files every 5-10 minutes to an AFS volume which is remote
over a 100Mbit dedicated WAN (33ms ping times)

with cache size = 500,000
with chunksize = 19
rsync -qau -timeout=60 --contimeout=30 <blah blah>

first rsync after afs client restart shows 0.15 - 0.25 seconds to complete

second, and subsequent rsync shows in excess of 30 seconds. It never
recovers to below one second.

Tested some parms... Decreasing cache size to 50,000 AND decreasing
chunksize to 13 gives consistent performance below 0.3 seconds

Increase cache size to 500m with chunksize 13 and the same behavior occurs
as before.

This seems counter intuitive... the 100 or so files do not go over the
500,000 block cache size. They are fairly small (10's to 100's of
kilobytes). Why would increasing cache size impact performance Negatively
in such a case?


-- 
Timothy Balcer / IT Services
Telmate / San Francisco, CA
Direct / (415) 300-4313
Customer Service / (800) 205-5510

--089e013a07d01501f504dfb264d5
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div><div><div><div><div><div><div><div>I have a tiny head=
 scratcher...<br><br>Tested this case:<br><br></div>rsync 100 or so files e=
very 5-10 minutes to an AFS volume which is remote over a 100Mbit dedicated=
 WAN (33ms ping times)<br>
<br></div>with cache size =3D 500,000<br></div>with chunksize =3D 19<br></d=
iv><div>rsync -qau -timeout=3D60 --contimeout=3D30 &lt;blah blah&gt;<br><br=
></div>first rsync after afs client restart shows 0.15 - 0.25 seconds to co=
mplete<br>
<br></div>second, and subsequent rsync shows in excess of 30 seconds. It ne=
ver recovers to below one second.<br><br></div>Tested some parms... Decreas=
ing cache size to 50,000 AND decreasing chunksize to 13 gives consistent pe=
rformance below 0.3 seconds<br>
<br></div>Increase cache size to 500m with chunksize 13 and the same behavi=
or occurs as before.<br><br></div>This seems counter intuitive... the 100 o=
r so files do not go over the 500,000 block cache size. They are fairly sma=
ll (10&#39;s to 100&#39;s of kilobytes). Why would increasing cache size im=
pact performance Negatively in such a case?<br>
<div><br><br><div><div><div><div><div><div><div><div><div><div>-- <br><span=
 style=3D"border-collapse:collapse;color:rgb(102,102,102);font-family:verda=
na,sans-serif;font-size:x-small">Timothy Balcer / IT Services<br>Telmate / =
San Francisco, CA<br>
Direct / </span><span style=3D"border-collapse:collapse;font-family:verdana=
,sans-serif;font-size:x-small"><font color=3D"#1155cc">(415) 300-4313</font=
><br><font color=3D"#666666">Customer Service /=A0</font><a value=3D"+18002=
055510" style=3D"color:rgb(17,85,204)">(800) 205-5510</a></span>
</div></div></div></div></div></div></div></div></div></div></div></div>

--089e013a07d01501f504dfb264d5--