<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Re: bufpages in Operating System - HP-UX</title>
    <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431877#M3361</link>
    <description>Terry,&lt;BR /&gt;since HP-UX can only make use of up to 400MB of buffer cache (if I&lt;BR /&gt;remember right), do NOT use "dbc_mac_pct", but use "bufpages" and&lt;BR /&gt;"nbuf" to allocate only that much RAM for it. You are wasting it, right now!&lt;BR /&gt;</description>
    <pubDate>Thu, 20 Jul 2000 22:35:34 GMT</pubDate>
    <dc:creator>Wodisch</dc:creator>
    <dc:date>2000-07-20T22:35:34Z</dc:date>
    <item>
      <title>bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431869#M3353</link>
      <description>Dynamic buffer cache allocation is enabled if the bufpages kernel parameter is set to zero.  I'm a little confused as to how much memory is allocated to buffer cache and how it is determined.  I have a couple of servers with similar configurations, but with totally different buffer cache amounts.  I'll include what the current settings are and hopefully someone can shed some light to as why they are so different.&lt;BR /&gt;&lt;BR /&gt;server 1:  (Makes sense to me)&lt;BR /&gt;kernel parameters&lt;BR /&gt;nbuf = 0&lt;BR /&gt;bufpages = 0&lt;BR /&gt;dbc_max_pct = 40&lt;BR /&gt;&lt;BR /&gt;Total Physical Memory = 1.50 GB&lt;BR /&gt;&lt;BR /&gt;Buffer Cache = 614  &lt;BR /&gt;&lt;BR /&gt;server 1:  (Doesn't Makes sense to me)&lt;BR /&gt;kernel parameters&lt;BR /&gt;nbuf = 0&lt;BR /&gt;bufpages = 0&lt;BR /&gt;dbc_max_pct = 40&lt;BR /&gt;&lt;BR /&gt;Total Physical Memory = 1.25 GB&lt;BR /&gt;&lt;BR /&gt;Buffer Cache = 159 MB&lt;BR /&gt;&lt;BR /&gt;Why the huge difference in Buffer Cache?  &lt;BR /&gt;&lt;BR /&gt;</description>
      <pubDate>Thu, 20 Jul 2000 11:22:38 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431869#M3353</guid>
      <dc:creator>Terry Gibbar</dc:creator>
      <dc:date>2000-07-20T11:22:38Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431870#M3354</link>
      <description>Buffer cache only gets used as needed the kernel parameter dbc_max_pct = 40 define the max. percentage of memory that can be used for buffer cache. The amount is NOT set aside initially, it grows as required. The buffer cache will also vary between dbc_min_pct &amp;amp; dbc_max_pct.&lt;BR /&gt;&lt;BR /&gt;Brian&lt;BR /&gt;&amp;lt;*(((&amp;gt;&amp;lt; er&lt;BR /&gt;</description>
      <pubDate>Thu, 20 Jul 2000 11:23:49 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431870#M3354</guid>
      <dc:creator>Brian M. Fisher</dc:creator>
      <dc:date>2000-07-20T11:23:49Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431871#M3355</link>
      <description>As it's name implies, Dynamic Buffer Cache can change dependant upon system load.&lt;BR /&gt;On the first node, the buffer cache is at 40% of 1.5Gb, or 600 Mb  give or take a few Mb.s&lt;BR /&gt;On the second node the setting allows for 40% of 1.25Gb or 500Mb, if it needs it!&lt;BR /&gt;&lt;BR /&gt;HtH</description>
      <pubDate>Thu, 20 Jul 2000 11:26:51 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431871#M3355</guid>
      <dc:creator>melvyn burnard</dc:creator>
      <dc:date>2000-07-20T11:26:51Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431872#M3356</link>
      <description>Terry, &lt;BR /&gt;As the setting of nbuf and bufpages have been set to zero,this causes the dynamic function of buffer cache paging increasing from the set minimum of 5% to set maximum of 40%. The second server's buffer cache is yet to grow to the maximum of 40% of Physical memory. It is behaving fine.&lt;BR /&gt;&lt;BR /&gt;Cheers!&lt;BR /&gt;</description>
      <pubDate>Thu, 20 Jul 2000 11:37:24 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431872#M3356</guid>
      <dc:creator>CHRIS_ANORUO</dc:creator>
      <dc:date>2000-07-20T11:37:24Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431873#M3357</link>
      <description>Thanks guys.....  Please let me know if the following makes sense.&lt;BR /&gt;&lt;BR /&gt;In the case of Node 1, the server has been up for 32 days and therefor the system has requested more memory for buffer cache.&lt;BR /&gt;&lt;BR /&gt;In the case of Node 2, the server has been up for 2 days and therefor the system has not requested much memory for buffer cache.&lt;BR /&gt;&lt;BR /&gt;One additional question:&lt;BR /&gt;1)  If free memory is running low, does the system take back from the memory that has been allocated for buffer cache?</description>
      <pubDate>Thu, 20 Jul 2000 12:15:48 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431873#M3357</guid>
      <dc:creator>Terry Gibbar</dc:creator>
      <dc:date>2000-07-20T12:15:48Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431874#M3358</link>
      <description>Terry, that's about right. If you've done a backup, that will usually increase the buffer cache quite a lot.&lt;BR /&gt;&lt;BR /&gt;As freemem reduces, the buffer cache will get smaller. However, freemem as to go below about 7MB for that to start.</description>
      <pubDate>Thu, 20 Jul 2000 12:27:13 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431874#M3358</guid>
      <dc:creator>Andy Monks</dc:creator>
      <dc:date>2000-07-20T12:27:13Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431875#M3359</link>
      <description>to all intents and purposes yes you ar ecorretc.&lt;BR /&gt;The answer to the question is that if the requirement for buffer cache drops, it will be released back to free memory and the system will then use it.&lt;BR /&gt;However, If the buffer cache decides it needs the space, then it does not get released, and you could end up with performance issues due to lack of free memory.&lt;BR /&gt;HTH&lt;BR /&gt;</description>
      <pubDate>Thu, 20 Jul 2000 12:28:49 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431875#M3359</guid>
      <dc:creator>melvyn burnard</dc:creator>
      <dc:date>2000-07-20T12:28:49Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431876#M3360</link>
      <description>Yes, it takes it back. If memory is full and the buffer cache is large, this will most likely cause paging, since the buffer cache typically shrinks slower than the rate at which new memory is needed.</description>
      <pubDate>Thu, 20 Jul 2000 12:33:28 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431876#M3360</guid>
      <dc:creator>CHRIS_ANORUO</dc:creator>
      <dc:date>2000-07-20T12:33:28Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431877#M3361</link>
      <description>Terry,&lt;BR /&gt;since HP-UX can only make use of up to 400MB of buffer cache (if I&lt;BR /&gt;remember right), do NOT use "dbc_mac_pct", but use "bufpages" and&lt;BR /&gt;"nbuf" to allocate only that much RAM for it. You are wasting it, right now!&lt;BR /&gt;</description>
      <pubDate>Thu, 20 Jul 2000 22:35:34 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431877#M3361</guid>
      <dc:creator>Wodisch</dc:creator>
      <dc:date>2000-07-20T22:35:34Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431878#M3362</link>
      <description>The information in the last post re 400Mb max buffer cache is incorrect.&lt;BR /&gt;If a system has 4Gb Ram, and the dbc_pct_max setting is left at default, the system can use up to 2Gb of RAM for buffer cache. This does present some performance issues unfortunately, but it will use the 2Gb if it needs to.</description>
      <pubDate>Fri, 21 Jul 2000 06:24:36 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431878#M3362</guid>
      <dc:creator>melvyn burnard</dc:creator>
      <dc:date>2000-07-21T06:24:36Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431879#M3363</link>
      <description>you can also force the buffer cache to be a fixed value by setting&lt;BR /&gt;dbc_max_pct = dbc_min_pct, then you can benchmark your system and&lt;BR /&gt;verify what settings provide the best performance on your system with your applications.</description>
      <pubDate>Fri, 21 Jul 2000 14:26:03 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431879#M3363</guid>
      <dc:creator>Jim Welch</dc:creator>
      <dc:date>2000-07-21T14:26:03Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431880#M3364</link>
      <description>Following on from this discussion. I have a server which has bufpages set and dbc parameters set. Does this mean that dynamic buferring is DISABLED and that the server will always reserve 785920 pages of memory for buffer cache regardless of the dbc parameters?&lt;BR /&gt;&lt;BR /&gt;bufpages                 785920 &lt;BR /&gt;nbuf                     608615 &lt;BR /&gt;ndilbuffers                  30 &lt;BR /&gt;dbc_max_pct                  50 &lt;BR /&gt;dbc_min_pct                   5&lt;BR /&gt;</description>
      <pubDate>Tue, 09 Jul 2002 22:20:09 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431880#M3364</guid>
      <dc:creator>Andrew Sobb</dc:creator>
      <dc:date>2002-07-09T22:20:09Z</dc:date>
    </item>
    <item>
      <title>Re: bufpages</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431881#M3365</link>
      <description>Regarding my previous question. Further investigation has revealed that the bufpages and nbuf parameters are set to zero but "sysdef | grep buf" has returned bufpages as the approximate value of the dbc_max_pct of the total memory in the host.&lt;BR /&gt;&lt;BR /&gt;Andrew</description>
      <pubDate>Tue, 09 Jul 2002 23:24:12 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/bufpages/m-p/2431881#M3365</guid>
      <dc:creator>Andrew Sobb</dc:creator>
      <dc:date>2002-07-09T23:24:12Z</dc:date>
    </item>
  </channel>
</rss>

