Solaris 8 cyclical page cache

I have read Sun's article about Solaris 8 cyclical page cache, but still don't understand how it works.
It did not mention how does kernel devide memory between the IO page buffer cache and applications.
Is it is first come, first served basis?
Is there any way to size the IO page buffer cache?
When app needs a page, and there are none any available, does it take it from IO buffer cache?
Does it mean that IO page cache will be 0 in size if application requires all physical memory?
In short, what is the algorithm?
I see me database application swap the executable/data parts heavily. It looks like it is due to kernel page cache competing for memory with the app. So, the problem is still not fixed after introduction of "priority paging" in Solaris 2.7 and "cyclical page cache" in Solaris 8?
Thanks for any feedback.

It will gradually use all memory except for a meg or so as page cache.
Thats what solaris does, if theres any memory thats not being used for actual work, it figures it might as well use it as page cache.
But the page cache will be released immediately if anything else wants the memory.
So you can think of the page cache pretty much as free memory. If fact most solaris tool simply report the page cache as part of the free memory.
So why would you want to limit the page cache?

Similar Messages

  • Solaris page cache limit

    Hello all,
    on a Solaris 10 (7/08) system we have here, we see the page cache increase over time. Does anyone know until what level this will increase and how to force it not to exceed a certain threshold?
    thanks,
    Twan

    It will gradually use all memory except for a meg or so as page cache.
    Thats what solaris does, if theres any memory thats not being used for actual work, it figures it might as well use it as page cache.
    But the page cache will be released immediately if anything else wants the memory.
    So you can think of the page cache pretty much as free memory. If fact most solaris tool simply report the page cache as part of the free memory.
    So why would you want to limit the page cache?

  • Clear the page cache

    Hi,
    We have recently started to develop our application on Solaris, before each regression test is run we need to clear the page cache so that performance is consistent. On Redhat/CentOS there is the following command:
    To free pagecache, dentries and inodes:
    echo 3 > /proc/sys/vm/drop_caches
    However I cant find an equivalent command for Solaris. We have been using tail -f /dev/zero to run a huge process until the memory has been flushed which isn't very reliable and slow. Anyone know of a different way to do this?

    If we are running multiple tests one after the other then the first test may make calls which causes information to be cached, then when the second test runs it may make use of the previously used pages in cache. This makes the second test results inconsistent as if it ran first the timing results would be different, also running through a debugger or with truss doesn't give the accuracy for timing that is required in this situation.

  • Clear page cache every time page is opened

    I have a menu page in Apex that points to 10 report pages. However each page retains the values in parameters. How can I clear the page cache every time I go from menu to individual report page? I see the CLEAR_PAGE_CACHE procedure but don't know where to place the call. I guess I haven't understood the event model of the Apex completely yet (I am from the Oracle Forms background).
    Jay

    Jay,
    You don't say how these links on the menu page are constructed. If you are using an apex list then for each list entry there would be a target page section in which you would put the target page ID in the Clear Cache field. For example, if the link goes to a page in the current application, say page 10, then you'd put 10 in the Clear Cache field.
    If you are building a link yourself, just put 10 (for page 10) in the clear-cache position of the f?p URL, e.g.,
    f?p=&APP_ID.:10:&APP_SESSION.::NO:10:P10_ITEM:&SOME_ITEM.
    Scott

  • Solaris Name Service Cache and Directory Proxy Problem

    We have some Solaris 10 clients ldapcliented to a Directory Proxy Server. After 15 minutes, the Solaris name service cache will fail to communicate to the proxy instance and the proxy instance's readconnectionsrefused attribute will start incrementing.
    At first it seemed we would need to increase the worker-threads and num-bind-limit, but those do not fix the problem.
    At the same time the name-service-cache starts failing, I am still able to query and search the proxy. I have set up a Jmeter test which continues to run and they never fail.
    It seems very consistent that the problem with the name-service-cache occurs every 15 minutes and I am able to reproduce this at the client's site and in my lab. Restarting either the proxy or the name-service-cache clears the problem.
    Has anyone else seen this problem?
    Edited by: 957466 on Sep 6, 2012 9:11 AM

    The idle-timeout on DSEE was set to none, which I believe is the default. I tried setting it to 1200 and 2400 seconds without success.
    h3. get-ldap-data-source-pool-prop
    <pre>
    client-affinity-bind-dn-filters : any
    client-affinity-criteria : connection
    client-affinity-ip-address-filters : any
    client-affinity-policy : write-affinity-after-write
    client-affinity-timeout : 20s
    description : -
    enable-client-affinity : false
    load-balancing-algorithm : proportional
    minimum-total-weight : 100
    proportion : 100
    sample-size : 100
    </pre>
    h3. get-ldap-data-source-prop
    <pre>
    bind-dn : none
    bind-pwd : none
    client-cred-mode : use-client-identity
    connect-timeout : 10s
    description : -
    down-monitoring-interval : inherited
    is-enabled : true
    is-read-only : false
    ldap-address : localhost
    ldap-port : ldap
    ldaps-port : ldaps
    monitoring-bind-dn : none
    monitoring-bind-pwd : none
    monitoring-bind-timeout : 5s
    monitoring-entry-dn : ""
    monitoring-entry-timeout : 5s
    monitoring-inactivity-timeout : 2m
    monitoring-interval : 30s
    monitoring-mode : proactive
    monitoring-retry-count : 3
    monitoring-search-filter : (objectClass=*)
    monitoring-search-scope : base
    num-bind-incr : 10
    num-bind-init : 2
    num-bind-limit : 1024
    num-read-incr : 10
    num-read-init : 2
    num-read-limit : 1024
    num-write-incr : 10
    num-write-init : 2
    num-write-limit : 1024
    proxied-auth-use-v1 : false
    ssl-policy : never
    use-read-connections-for-writes : false
    use-tcp-keep-alive : true
    use-tcp-no-delay : true
    </pre>

  • Report (not Page) caching to improve report loading times

    We are trying out Cystal Reports Server 2008 as a replacement for Crystal Reports CR XI R2 for an ASP.NET Web Application. Running some tests we found that the same report loaded from the server was far slower than loading the same report locally, using the CRXI SDK.
    Being the Clone/Refresh strategy on the standalone SDK several times faster than the reportAppFactory.OpenDocument on the server platform, i think that the server is loading from scratch the report each time I execute the OpenDocument method. So what I am seraching for is, Is there a way to setup or tweak report caching on Crystal Server 2008? What I need is not page caching, but the whole report, which is used different databases each time it's executed (implying different data for each execution).
    Some insight on the problem:
    As I posted on a previous thread we are moving from CR XI due to a limit on the number of active instances suported by the SDK (74 instances). We reach this limit because of a workaround for improving report execution times, leaving instances of each report active, so the next execution just recicles the active instance, greatly reducing the overall process time. Therefor introducing Crystal Server is a setback on this module (due to it's actual performance).
    Part of the problem resides on the complexity of the report. Using the standalone SDK the report takes several seconds to load from disk, and several seconds to execute. Regretably, we are not at liberty to change the reports' structure, so optimizing it is beyond posibility for now.
    Thanks in advance,
    Gustavo

    Hello Maggie,
    >> How can I get an all-encoompassing CSV file of the main report (Page 2014)?
    This might be possible using the advanced print server configuration, with BI Publisher, using the same technique that is being used to print master-details reports (which is a type of a multi-region report) - http://www.oracle.com/technology/products/database/application_express/html/configure_printing.html . The standard print server configuration only supports reports with a single region. If you have BIP in your organization, that’s great. Otherwise, CSV files don’t warrant it.
    The only other option, I can see, is to create the CSV file manually, using the technique described in the following Blog entry, by Scott Spendolini - http://spendolini.blogspot.com/2006/04/custom-export-to-csv.html .
    Regards,
    Arie.

  • Page Zero (0) - Regions Caching / Page Caching not available

    Hi,
    My page zero display information from a calendar table (fiscal year, etc.). This information will be the same all day long (get updated at midnight). It's not possible to enable caching of regions on page zero. It's also impossible to enable caching of the page zero.
    Why?
    Thanks,
    Louis-Guillaume
    [My Blog|http://lgcarrier.blogspot.com]
    [APEX and jQuery demos|http://www.insum.ca/jquery]

    Louis-Guillaume,
    Good question. We have asked ourselves this and are considering removing the restriction in 4.0. I think it initially had to do with preventing unwanted side effects and anomalies although none of us can recall the details right now. We will have to carefully look at ways to make this something developers can use easily while preventing strange results. First of all, caching of page 0 doesn't make sense. You want to be able to cache regions on page 0. Now, say your page 0 has two regions, P0_CACHED and P0_DYN (one cached, one dynamic). And your page 10 has region P10_CACHED. When page 10 is rendered, you'll get:
    P0_CACHED
    P0_DYN
    P10_CACHED
    That looks okay, you get one cached region from page 0, one dynamic region from page 0, and one cached region from page 10. Of course you have controls with which to purge any individual region from the cache so you can cause P0_CACHED to be refreshed whenever you like.
    But say page 20 is a cached page. When it renders you'll get:
    P0_CACHED <-- not from the page 0 region cache but from the page 20 page cache, regardless of the "stale-ness" of the page 0 cache
    P0_DYN <-- not rendered dynamically from page 0 but retrieved as part of the page 20 page cache. But this content may be different from the content for the same page 0 region displayed on page 10 one second ago or one second from now.
    P20_DYN <-- from the page cache as is normal for a dynamic region on a cached page
    Hardly a thorough treatment, I realize. Just wanted you to know some of the aspects we have to consider. We will also have to read this thread again to come back up to speed: V3 Caching - any more info? .
    Scott

  • Problem with page cache clearing

    I have a button that exits HTML DB and returns to our portal main page. That is done through a URL, and not a page number. There is no way to reset page cache with a URL branch. So I made a page that has a branch that calls the URL, and then I changed the button on the page to branch to this page and do the proper page cache resets before calling the other page that branches to the URL. However, the page cache is not being reset when I do this.
    Here is the page flow
    page 10 - Select a record for update
    page 20 - Display data using an ID that is on page 10
    If Return to Menu is clicked branch to page 1 and reset page 10 and page 20
    Page 1 - Branches to URL
    When I go back to page 10 from the application server page page 10 & 20 are still populated.
    If I change page 20's return to menu to branch to page 10, then the reset does occur. What am I doing wrong? How do I reset the pages when leaving HTML DB?

    Dwayne,
    Validations, like other components, can have conditions that observe the current request value. This value is set to the button name when a button submits the page. So you could have a condition on a validation like 'When request != Expression 1' and in the Expression 1 field type the button name (check the button attributes to confirm the case-sensitive name and submit the page in debug mode to see the value of request when you click the button).
    About clearing page cache when branching to URL -- not sure what you're after: Cache can be cleared if an HTML DB page is invoked via URL. You specify clear-cache options in the branch definition which you can observe in the URL that invokes the target page. If you mean that you want the cache of one or more HTML DB pages to be cleared "when" a branch to a non-HTML DB page is taken, that doesn't really make sense because nothing much happens during a branch except to call the HTML DB engine's accept procedure or to do a URL redirect. So all you can do is specify some action to happen "before" a branch is taken, e.g., clear cache on pages x,y,z using a page process, or specify that the clear cache action is to occur "after" the branch is taken as manifested in the resulting URL to an HTML DB page. For something to happen "when" a branch is being done presumably would be after page processing but before a redirect, which we already support in the form of page processes.
    Scott

  • Image from BLOB now showed when page cache set

    Hello,
    I have a page that display images from BLOB,
    After setting page cache to 1hour, page load faster but images not showed.
    Please help me on this.
    Thank you.

    Hi Halit,
    For me i dont thing, this problem occurred. As i checked by setting the page cache to 1 hour. After that images loaded normally as usual.
    I tried with my mozilla browser. In which browser you tried, may be your browser fault.
    Kindly check it.
    Brgds,
    Mini
    Mark Answers Promptly

  • Partial Page Caching in 902

    What, if Any, partial page caching using ESI tags can be done in Portal V2 or with the PDK?

    There is no PPC available through Portal in 902.
    PPC will be available in 903/4.
    Obviously you can still use PPC in your on code that directly accesses Webcache
    Thanks
    -JP

  • Putting a page cache as old

    Hi, we are caching page definition & content, but at sometimes
    we want to force a recalculation programmatically, i.e., to put
    the page cache as old.
    I tried to used the wwutl_api_cache package, but it (silently)
    doesn't change anything. We are using Portal 3.0.9.8.
    Any clues?
    Thanks in advance.

    > When I press F12 to preview the site, the link to the
    Orient page is as
    > follows:
    >
    > C:\Documents and Settings\Tracy\My
    Documents\Serenity\Orient
    > But how do I make it so the .htm is there and doesn't
    need to be typed in
    > order to preview the page? I have this problem with two
    different pages.
    okay-- so the path is wrong.
    Open the page for editing in dreamweaver.
    Select that link with the mouse.
    In the Properties Inspector, use the folder icon next to the
    Link box to
    browse to the correct file.
    Alan
    Adobe Community Expert, dreamweaver
    http://www.adobe.com/communities/experts/

  • How to disable a JSP page caching?

              Hello there,
              How do I can disable a JSP page caching? This is becoz when the first time i ran
              a EJB calling from a JSP, it works, but after the second time, it unable to call
              the same EJB again, I double check the console and the database, the same records
              does not appears after the second calling, i suspect is caching problem.
              TQ
              Neo
              KL
              

    Neo, wls does not cache jsp page by default. We do have a custom tag for
              it as you can see here
              http://e-docs.bea.com/wls/docs61/jsp/customtags.html#overview.
              Regards,
              Joseph Nguyen
              BEA Support
              "Neo" <[email protected]> wrote in message
              news:3d4a428d$[email protected]..
              >
              > Hello there,
              >
              > How do I can disable a JSP page caching? This is becoz when the first time
              i ran
              > a EJB calling from a JSP, it works, but after the second time, it unable
              to call
              > the same EJB again, I double check the console and the database, the same
              records
              > does not appears after the second calling, i suspect is caching problem.
              >
              > TQ
              >
              > Neo
              > KL
              

  • Invalidating portal page cache

    On a portal page, contents are rendered by portlets using java programs. With page caching on, the entire portal page is cached. Performance is great when users are navigating back to this portal page from other pages.
    Because the portlets hold session context in the application, this portal page is invalid once a user signs out or closes the browser window. The problem is that if this portal page is cached, when the user signs in again portal would present the cached page without going to the portlets to re-establish a new session context.
    What can we do to invalidate the cache of this portal page? We found nothing in the portal to specify any invalidation rules for a portal page. We believe we cannot invalidate the cached portal page from the application.
    Any ideas will be appreciated,
    Chung

    With the following settings:
    1. Portal page caching:
    Cache Page Definition Only
    2. Provider.xml of portlets:
    <pageExpires>20</pageExpires>
    3. Provider registration info:
    Require portal user specific session information. (Option checked)
    we get the following results:
    - When portal user navigates back to the portal page consisting the portlets,
    the page is rendered quickly. It's because provider.xml says portlets are good for 20 minutes.
    - When a user logs out and within 20 minutes logs in again IN A NEW BROWSER WINDOW, a new session id will be used, portal will not use the cached portlets. Hence, requests will be made to the application and a new application session context will be created.
    However, if the user uses the same browser window, then cached portlets will be used. It would encounter the application session timeout error. This restriction might be acceptable to customers.
    We will recommend the above settings to the customer. Please verify that this is a valid solution.

  • Page & cache size performance tuneup

    Hi
    I am doing performance evaluation on BDB. Please help me in find answer to below queries.
    1. page size: Do I need to give the page size based on my XML document size. Is there any relation(formula) between page size & XML document size to get optimum memory usage?
    2. cache size: Is cache size needs to be equal/more than the doc size to minimize the query response time? Could you please suggests a optimum cache size for 1MB XML document?
    3. I have stared with BDB version 2.3.10, but i read in this forum there is some performance improvement in BDB version 2.3.10. What version i should use for my evaluation? Is the latest(4.6.21) is best(stable)?
    4. Is any other parameters ( other than page & cache size) I need to tuneup to get optimum memory usage & minimal CPU utilization?
    Is there any reference document I can get more details on BDB performace?
    Thanks,
    Santhosh

    Hi Santhosh,
    It’s hard to give solid suggestions without knowing more about your application, what you are measuring and what your performance requirements are. What language are you implementing in?
    Is query response time most important, or document insertion or updates?
    I am going to request that you respond to this Performance Questionaire and answer as many questions as you can at this time. Send the questionaire to me at Ron dot Cohen at Oracle.
    http://forums.oracle.com/forums/ann.jspa?annID=426
    In addition to the information requested, you can see from the questionaire that the utility
    Db_stat –m is useful to look at a number of things including the effectiveness of the cache size you have.
    Have you taken any measurements yet? I would suggest going with the default pagesize but using a cachesize larger than the default. I don’t know how much real memory you have but for a first measurement you could try a cachesize of 100MB-500MB (or larger) depending on your workload and how much memory you have available. I am not recommending that as a final cache size, just giving you a number to start with.
    http://tinyurl.com/2mfn6f
    You will likely find a lot of improvements in performance can be obtained by your indexing strategy. This may be where you get the best results. You may want to spend some time reviewing that and the documentation on indexes:
    http://tinyurl.com/2522sc
    Also, take a look in the same document at the indexing sections.
    Berkeley DB XML 2.3 (Berkeley DB 4.5.20) should be fine to start (though you may have read on this forum about the speed improvements in Berkeley DB XML 2.4 which is currently in test mode).
    Please do respond to the survey, send it to me and we will try to help you further.
    Ron

  • Difference between Page cache and cahelist in mdb output

    Hello
    When perfoming a "mdb -k" dee command ::memstat a got an output which summerize memory allocation and I cannot make the difference between the "page cache" and the "Free(cachelist)" rows. Both are used by file data by I can't make any difference.
    Does anyone know which differs between these two counter?
    Thank you

    Hi,
    Query Precalculation
    The distribution type Precalculated Value Set allows you to fill variables of the type precalculated value sets with values for characteristic values in the background. The precalculated value sets are then available as variable values in BEx queries.
    The following cases are typical examples of where the execution of precalculated value sets in the background is used:
    ●     When analyzing target groups, because you can specify a fixed value
    ●     When complex selections are involved such as with top N queries, used to greatly improve query performance
    ●     Instead of variables that are filled from a query event
    http://help.sap.com/saphelp_nw70ehp1core/helpdata/en/2a/b37f425b595551e10000000a1550b0/frameset.htm
    Filling OLAP Cache
    The distribution type Fill OLAP Cache allows you to precalculate queries and to fill the OLAP cache with the generated data.
    If the users call Web Applications, queries, reports or workbooks that are based on this data, the access time is significantly reduced and the workload on the application server is considerably less.
    http://help.sap.com/saphelp_nw70ehp1core/helpdata/en/2a/b37f425b595551e10000000a1550b0/frameset.htm
    Hope this helps
    Regards
    Snehith

Maybe you are looking for

  • URGENT.PLEASE HELP!!!Print graphics problem

    Hi, Please help me in this one.I try to print Graphics but I can't. 1) Spooling lasts too long 2) After spooling , printing does not continue 3) An empty page is printed instead 4) Printing process takes about 60 Mbytes of RAM to perform (??!!!) Here

  • Annoying little chapter glitch

    Okay, I have two seperate movies on one menu with no chapter markers. If I'm in movie #1, and press the chapter advance button on the remote, it automatically goes to movie #2. In movie #2, the chapter reverse button doesn't do anything, but if I pre

  • Can't download adobe flash

    I need the new version of flash on my macbook pro but it will not download. I tried to download google chrome instead and it will not download either. But this time there was a small error message in the middle of the screen that says Plug-in Failure

  • Audio Interface Compatibility w/ Soundtrack Pro

    I am a fairly new Soundtrack Pro user but have used other multitrack audio editing programs in the past. I am in the market for an audio interface to run my new studio monitors and mics etc. The audio interface I am considering buying is the M-AUDIO

  • Dropdown at selection screen field

    Hi folks, I know how to create drop down enabled field at selection screen but my problem is how to provide the input fie name or unix file name to this list.. file type will be DAT please help me out thanks in advance. Edited by: KARTIKEY SINGH on N