Impressive bit of research R.Gunther :)<div><br></div><div>Cheers</div><div><br><br><div class="gmail_quote">On Sat, Sep 29, 2012 at 6:36 PM, R.Gunther <span dir="ltr"><<a href="mailto:rigun@rigutech.nl" target="_blank">rigun@rigutech.nl</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
  
    
  
  <div bgcolor="#FFFFFF" text="#000000">
    <div>=====================================<br>
      I switched back to linux with my osgrid region.<br>
      I have some strange sideeffect on my train megasim, as soo the
      train is rideing and am on the sim it seems to crash pretty fast <br>
      with a nice stacktrace and above that some slow requests on
      serveral parts.<br>
      <br>
      22:16:02 - [BASE HTTP SERVER]: Slow handling of 498 POST /CAPS/
      ......  took 5708ms<br>
      22:16:05 - [BASE HTTP SERVER]: Slow handling of 505 POST /CAPS/
      ...... took 4644ms<br>
      22:17:35 - [FORMS]: Slow request 45 POST ........ took 100125ms,<br>
      23:19:17 - [FORMS]: Slow request 138 POST.......  took 100024ms,<br>
      23:19:58 - [USER AGENT CONNECTOR]: Unable to contact remote server
      ...... for GetServerURLs<br>
      Region (root) # Stacktrace:<br>
      <br>
      Native stacktrace:<br>
      <br>
              mono() [0x498a19]<br>
              mono() [0x4e999f]<br>
              mono() [0x41f617]<br>
              /lib64/libpthread.so.0(+0xf140) [0x7effa5a6c140]<br>
              mono() [0x5e04d9]<br>
              mono() [0x5dd4a0]<br>
              mono() [0x5e0cfc]<br>
              mono() [0x5e15d2]<br>
              mono() [0x5e1c07]<br>
              mono() [0x5e263d]<br>
              mono() [0x5dc292]<br>
              mono() [0x5dc575]<br>
              mono() [0x5dc7ee]<br>
              mono(mono_array_new_specific+0xe2) [0x590602]<br>
              [0x4186fd7d]<br>
      <br>
      Debug info from gdb:<br>
      <br>
      So have done a bit searching, thereare  maby a few hints.<big><br>
        <small>I try to explain it best i can, hope 1 of them is
          direction to the solution.<br>
          Its maby more a collection of ideas , more the needle in the
          hay.<br>
          Not C# expert, but what i see the request code is pretty
          basic.<br>
          The one opensim uses looks a bit like whats used on <br>
        </small></big><big><small><big><small>**********</small></big><br>
        </small></big><br>
      <big><small><big><small><a href="http://www.codeproject.com/Articles/13486/A-Simple-Crawler-Using-C-Sockets#WebRequest" target="_blank">http://www.codeproject.com/Articles/13486/A-Simple-Crawler-Using-C-Sockets#WebRequest</a></small></big><br>

          and look at the section, "2. <b>WebRequest and WebResponse
            problems:"  </b>am quoting a piece of text from that
          website<br>
          <br>
          ""This code works well but it has a very serious problem as
          the <code>WebRequest</code> class function <code>GetResponse</code>
          locks the access to all other processes, the <code>WebRequest</code>
          tells the retrieved response as closed, as in the last line in
          the previous code. So I noticed that always only one thread is
          downloading while others are waiting to <code>GetResponse</code>.
          To solve this serious problem, I implemented my two classes, <b><code>MyWebRequest</code></b>
          and <b><code>MyWebResponse</code></b>.""<br>
        </small><small><br>
          **********<br>
          HttpWebRequest.GetResponse Method</small></big> seems to
      block.<br>
      <br>
      the Microsoft page say the following:
<a href="http://msdn.microsoft.com/en-us/library/system.net.httpwebrequest.getresponse%28v=vs.71%29.aspx" target="_blank">http://msdn.microsoft.com/en-us/library/system.net.httpwebrequest.getresponse%28v=vs.71%29.aspx</a><br>

      ""This method blocks waiting for content to post; if there is no
      time-out set and you do not provide content, your application will
      block indefinitely.""<br>
      <br>
      <big><small><big><small>**********</small></big></small></big><br>
      <br>
      Maby the ServicePointManager.Expect100Continue Property is
      intressting to use ? 
<a href="http://msdn.microsoft.com/en-us/library/system.net.servicepointmanager.expect100continue%28v=vs.71%29.aspx" target="_blank">http://msdn.microsoft.com/en-us/library/system.net.servicepointmanager.expect100continue%28v=vs.71%29.aspx</a><br>

      <br>
      <big><small>**********</small></big><br>
      last thing that i want to point to is this site
<a href="http://stackoverflow.com/questions/2519655/httpwebrequest-is-extremely-slow" target="_blank">http://stackoverflow.com/questions/2519655/httpwebrequest-is-extremely-slow</a><br>
      The talk about that the http proxy is enabled default, and that
      disable it can speed up the process.<br>
      Most of the above things come from this page.<br>
      <br>
      =====================================<div><div class="h5"><br>
      <br>
      On 2012-09-22 02:22, Justin Clark-Casey wrote:<br>
    </div></div></div><div><div class="h5">
    <blockquote type="cite">There
      are no timeouts on these particular requests, which probably isn't
      a good thing - OpenSimulator is inconsistent when it comes to
      timeouts.  However, I suspect that's oging to be peripheral to the
      issue here.
      <br>
      <br>
      On 20/09/12 11:44, M.E. Verhagen wrote:
      <br>
      <blockquote type="cite">I wonder if there is a timeout on the
        fetch request ?
        <br>
        Or are they stacking up and just waiting till they are in the
        front of the line to be processed ?
        <br>
        <br>
        2012/9/20 Justin Clark-Casey <<a href="mailto:jjustincc@googlemail.com" target="_blank">jjustincc@googlemail.com</a>
        <a href="mailto:jjustincc@googlemail.com" target="_blank"><mailto:jjustincc@googlemail.com></a>>
        <br>
        <br>
            Fetch times of 1068560ms (which is more than 15 minutes!)
        are pathological - something is wrong.
        <br>
        <br>
            In the first case I would suspect an issue with the
        connection to the OSGrid inventory service.  Either it is being
        <br>
            overloaded (OSGrid actually runs multiple copies of the
        ROBUST inventory service and load-balances between them
        <br>
            using Nginx, but perhaps the limits are being hit), things
        are being massive lagged because of lots of requests
        <br>
            hitting the .net default connection limit, as Melanie
        discussed, or you have a network issue.
        <br>
        <br>
            However, I'm a little surprised that the OSGrid Inventory
        Service request within FetchInventoryDescendents2 isn't
        <br>
            making it's own complaint about a long request.  The log
        message you're seeing does cover the whole of the CAPs call
        <br>
            so I suppose there is some chance there's some other cause,
        though it is a pretty simple capability.
        <br>
        <br>
            If you want to see if viewers are doing a lot of inventory
        fetches, then you can try the command "debug http 3" on
        <br>
            the simulator console.  This will show every incoming http
        request from viewers (and other simulators) and the HTTP
        <br>
            URI they are hitting.  Very recent code also has levels 4
        and 5 which will log a snippet of the incoming request and
        <br>
            the entire request, though you probably don't want to use
        them for any length of time.
        <br>
        <br>
        <br>
            On 14/09/12 11:33, Ai Austin wrote:
        <br>
        <br>
        <br>
                    From: Melanie <<a href="mailto:melanie@t-data.com" target="_blank">melanie@t-data.com</a>
        <a href="mailto:melanie@t-data.com" target="_blank"><mailto:melanie@t-data.com></a>>
        <br>
        <br>
                    You're using a viewer that fetches the user
        inventory in the
        <br>
                    background on login. ...
        <br>
                    You will need to disable inventory fetch on startup
        and/or
        <br>
                    viewerside AO.
        <br>
        <br>
        <br>
                Thanks for the explanation Melanie... that sounds like
        it is the type of problem I am seeing on our add on OSGrid
        <br>
                regions... but not on our own Openvue grid whcih
        essentially runs on identical hardware and with identical
        <br>
                network setups.
        <br>
        <br>
                I am observing the problem with both LL 3.3.4 and
        Firestorm/Pheonix 4.2.2...
        <br>
        <br>
                I don't have the viewer provided sAO turned on in
        Firestorm,  but I just looked at all the settings I can find in
        LL
        <br>
                3.3.4 and Firestorm 4.2.2 and cannot see one that refers
        to http fetching via http.  Although I am sure I have
        <br>
                seen this
        <br>
                in the past on earlier viewers.  Which Preferences tab
        is it on?
        <br>
        <br>
                    From: James Stallings II
        <<a href="mailto:james.stallings@gmail.com" target="_blank">james.stallings@gmail.com</a>
        <a href="mailto:james.stallings@gmail.com" target="_blank"><mailto:james.stallings@gmail.com></a>>
        <br>
        <br>
                    I would recommend (assuming firestorm here) setting
        both inventory fetch
        <br>
                    and texture fetch to http in the settings, and
        making the following changes
        <br>
                    to opensim.ini:
        <br>
                    [ClientStack.LindenCaps]
        <br>
                         ...
        <br>
                         Cap_GetTexture = "localhost"
        <br>
                         Cap_GetMesh = "localhost"
        <br>
                         ; This is disabled by default. Change if you
        see fit. Note that
        <br>
                         ; serving this cap from the simulators may lead
        to poor performace.
        <br>
                          Cap_FetchInventoryDescendents2 = "localhost"
        <br>
                          Cap_FetchInventory2 = "localhost"
        <br>
        <br>
        <br>
                Yep, that what I am set to already James.. I just
        checked. Thanks.
        <br>
                _________________________________________________
        <br>
                Opensim-users mailing list
        <br>
                <a href="mailto:Opensim-users@lists.berlios.de" target="_blank">Opensim-users@lists.berlios.de</a>
        <a href="mailto:Opensim-users@lists.berlios.de" target="_blank"><mailto:Opensim-users@lists.berlios.de></a>
        <br>
               
        <a href="https://lists.berlios.de/__mailman/listinfo/opensim-users" target="_blank">https://lists.berlios.de/__mailman/listinfo/opensim-users</a>
        <a href="https://lists.berlios.de/mailman/listinfo/opensim-users" target="_blank"><https://lists.berlios.de/mailman/listinfo/opensim-users></a>
        <br>
        <br>
        <br>
        <br>
            --
        <br>
            Justin Clark-Casey (justincc)
        <br>
            OSVW Consulting
        <br>
            <a href="http://justincc.org" target="_blank">http://justincc.org</a>
        <br>
            <a href="http://twitter.com/justincc" target="_blank">http://twitter.com/justincc</a>
        <br>
        <br>
            _________________________________________________
        <br>
            Opensim-users mailing list
        <br>
            <a href="mailto:Opensim-users@lists.berlios.de" target="_blank">Opensim-users@lists.berlios.de</a>
        <a href="mailto:Opensim-users@lists.berlios.de" target="_blank"><mailto:Opensim-users@lists.berlios.de></a>
        <br>
            <a href="https://lists.berlios.de/__mailman/listinfo/opensim-users" target="_blank">https://lists.berlios.de/__mailman/listinfo/opensim-users</a>
        <a href="https://lists.berlios.de/mailman/listinfo/opensim-users" target="_blank"><https://lists.berlios.de/mailman/listinfo/opensim-users></a>
        <br>
        <br>
        <br>
        <br>
        <br>
        --
        <br>
        Groningen en Hannover Opensims:
        <a>secondlife://meverhagen.nl:8002:Hannover</a> ZW/
        <br>
        <br>
        <br>
        _______________________________________________
        <br>
        Opensim-users mailing list
        <br>
        <a href="mailto:Opensim-users@lists.berlios.de" target="_blank">Opensim-users@lists.berlios.de</a>
        <br>
        <a href="https://lists.berlios.de/mailman/listinfo/opensim-users" target="_blank">https://lists.berlios.de/mailman/listinfo/opensim-users</a>
        <br>
        <br>
      </blockquote>
      <br>
      <br>
    </blockquote>
    <br>
  </div></div></div>

<br>_______________________________________________<br>
Opensim-users mailing list<br>
<a href="mailto:Opensim-users@lists.berlios.de">Opensim-users@lists.berlios.de</a><br>
<a href="https://lists.berlios.de/mailman/listinfo/opensim-users" target="_blank">https://lists.berlios.de/mailman/listinfo/opensim-users</a><br></blockquote></div><br><br clear="all"><div><br></div>-- <br>===================================<br>
<a href="http://simhost.com" target="_blank">http://simhost.com</a><br><a href="http://twitter.com/jstallings2" target="_blank">http://twitter.com/jstallings2</a><br><a href="http://www.linkedin.com/pub/5/770/a49" target="_blank">http://www.linkedin.com/pub/5/770/a49</a><br>

</div>