<div dir="ltr">Miguel<div><br></div><div>This was an error in the development version of CarpetIOHDF5 -- at one place, &quot;component&quot; should have been &quot;local_component&quot;. This has now been corrected.</div><div><br></div><div>-erik</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Aug 7, 2015 at 11:22 AM, Miguel Zilhão <span dir="ltr">&lt;<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a>&gt;</span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">hi Erik,<br>
<br>
actually, i&#39;ve just noticed that, for whatever reason, i was using Carpet&#39;s master branch... the rest of my ET installation does point to ET_2015_05 (and i thought Carpet did as well). i must have typed &#39;git checkout master&#39; in the wrong place at some point, and never noticed it.<br>
<br>
so the &#39;fix&#39; i mentioned in my previous email applies to Carpet&#39;s current master branch. if i checkout branch ET_2015_05 and compile that (as i thought i was doing), it works fine without me needing to change anything.<br>
<br>
thanks,<br>
Miguel<span class=""><br>
<br>
<br>
On 07/08/15 15:08, Erik Schnetter wrote:<br>
</span><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">
Miguel<br>
<br>
Thanks for tracking this down. I believe this code looks already different on the current master.<br>
Can you just disable it, as you did? The set of active grid points is important for some very new<br>
post-processing and visualization tools, but you probably don&#39;t need it, and you don&#39;t need it in a<br>
checkpoint file unless you want to visualize the data stored there.<br>
<br>
-erik<br>
<br>
<br></span><span class="">
On Fri, Aug 7, 2015 at 8:31 AM, Miguel Zilhão &lt;<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a> &lt;mailto:<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a>&gt;&gt; wrote:<br>
<br>
    Erik,<br>
<br>
    some more info. i briefly went through git history and found the offending commit. up to Carpet<br>
    commit d52042b867eea1e0770b6de87ff3929ab7b9d297 everything works fine for me. from commit<br>
    69c73fb12d2ee41f01928122cd178d1bae9f8e13 onward i get the error i mentioned.<br>
<br>
    indeed, in the current state, if i remove writing of the &quot;active&quot; attribute everything works<br>
    just fine:<br>
<br>
    diff --git a/CarpetIOHDF5/src/Output.cc b/CarpetIOHDF5/src/Output.cc<br>
    index ec9b090..30c017d 100644<br>
    --- a/CarpetIOHDF5/src/Output.cc<br>
    +++ b/CarpetIOHDF5/src/Output.cc<br>
    @@ -894,10 +894,11 @@ static int AddAttributes (const cGH *const cctkGH, const char *fullname,<br>
          HDF5_ERROR (H5Awrite (attr, H5T_NATIVE_INT, &amp;ioffset[0]));<br>
               HDF5_ERROR (H5Aclose (attr));<br>
<br>
    -    ostringstream buf;<br></span>
    -    buf &lt;&lt; (<a href="http://vdd.at" rel="noreferrer" target="_blank">vdd.at</a> &lt;<a href="http://vdd.at" rel="noreferrer" target="_blank">http://vdd.at</a>&gt;(Carpet::map)-&gt;<br>
    - <a href="http://local_boxes.at" rel="noreferrer" target="_blank">local_boxes.at</a> &lt;<a href="http://local_boxes.at" rel="noreferrer" target="_blank">http://local_boxes.at</a>&gt;(mglevel).at(refinementlevel).at(component).active);<span class=""><br>
    -    WriteAttribute(dataset, &quot;active&quot;, buf.str().c_str());<br>
    +  //   ostringstream buf;<br></span>
    +  //   buf &lt;&lt; (<a href="http://vdd.at" rel="noreferrer" target="_blank">vdd.at</a> &lt;<a href="http://vdd.at" rel="noreferrer" target="_blank">http://vdd.at</a>&gt;(Carpet::map)-&gt;<br>
    +  // <a href="http://local_boxes.at" rel="noreferrer" target="_blank">local_boxes.at</a> &lt;<a href="http://local_boxes.at" rel="noreferrer" target="_blank">http://local_boxes.at</a>&gt;(mglevel).at(refinementlevel).at(component).active);<div><div class="h5"><br>
    +  //   WriteAttribute(dataset, &quot;active&quot;, buf.str().c_str());<br>
    +<br>
        }<br>
<br>
        if (is_index) {<br>
<br>
<br>
    i can&#39;t tell why it breaks once that bit of code is in, though...<br>
<br>
    thanks,<br>
    Miguel<br>
<br>
<br>
<br>
    On 06/08/15 15:43, Miguel Zilhão wrote:<br>
<br>
        hi Erik,<br>
<br>
            Can you look at &lt;<a href="https://trac.einsteintoolkit.org/ticket/1800" rel="noreferrer" target="_blank">https://trac.einsteintoolkit.org/ticket/1800</a>&gt;? Can you try replacing<br>
            &quot;reflevel&quot; in<br>
            line 816 of the file CarpetIOHDF5/src/Output.cc with &quot;refinementlevel&quot;, and see whether<br>
            this avoids<br>
            the problem?<br>
<br>
<br>
        thanks, but this does not seem to help... i still get the same error and backtrace in gdb:<br>
<br>
        #6  0x00007ffff4f6d4d5 in std::__throw_out_of_range_fmt(char const*, ...) ()<br>
              from /usr/lib64/libstdc++.so.6<br>
        #7  0x0000000000c70402 in _M_range_check (__n=&lt;optimized out&gt;,<br>
               this=&lt;optimized out&gt;) at /usr/include/c++/5/bits/stl_vector.h:803<br>
        #8  at (__n=&lt;optimized out&gt;, this=&lt;optimized out&gt;)<br>
               at /usr/include/c++/5/bits/stl_vector.h:824<br>
        #9  CarpetIOHDF5::AddAttributes (cctkGH=cctkGH@entry=0x1b1b930,<br>
               fullname=fullname@entry=0x3e52d760 &quot;ML_BSSN::cA&quot;, vdim=3,<br>
               refinementlevel=refinementlevel@entry=0, request=request@entry=0x3df4c990,<br>
               bbox=..., dataset=83886080, is_index=false)<br>
               at<br>
        /home/mzilhao/Trabalho/projectos/ET/Cactus/arrangements/Carpet/CarpetIOHDF5/src/Output.cc:899<br>
        #10 0x0000000000c72e5a in CarpetIOHDF5::WriteVarChunkedParallel (<br>
               cctkGH=cctkGH@entry=0x1b1b930, outfile=outfile@entry=167<br>
<br>
<br>
        Miguel<br>
<br>
<br>
            On Wed, Aug 5, 2015 at 8:11 PM, Miguel Zilhão &lt;<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a><br></div></div><div><div class="h5">
            &lt;mailto:<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a>&gt; &lt;mailto:<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a> &lt;mailto:<a href="mailto:mzilhao@ffn.ub.es" target="_blank">mzilhao@ffn.ub.es</a>&gt;&gt;&gt; wrote:<br>
<br>
                  hi all,<br>
<br>
                  i&#39;m running latest ET Hilbert on openSUSE tumbleweed and i&#39;m having the following<br>
            issue. upon<br>
                  trying to run a simple head-on collision configuration with McLachlan (attached<br>
            parameter file),<br>
                  i get the error<br>
<br>
                     INFO (CarpetIOHDF5): ---------------------------------------------------------<br>
                     INFO (CarpetIOHDF5): Dumping initial checkpoint at iteration 0, simulation time 0<br>
                     INFO (CarpetIOHDF5): ---------------------------------------------------------<br>
                     terminate called after throwing an instance of &#39;std::out_of_range&#39;<br>
                       what():  vector::_M_range_check: __n (which is 1) &gt;= this-&gt;size() (which is 1)<br>
                       Rank 1 with PID 5958 received signal 6<br>
<br>
                  when writing the checkpoint file.<br>
                  this only happens if i run with more than one MPI process; with a single processor<br>
            it runs fine.<br>
<br>
                  i&#39;m compiling with gcc-5, but i find the same problem with gcc-4.8. i was running<br>
            this very same<br>
                  configuration just fine a couple of months ago, so it must have been some update<br>
            i&#39;ve made in<br>
                  the meantime (either to my OS or to ET).<br>
                  i&#39;ve also tried with different configurations and the outcome is the same.<br>
<br>
                  i&#39;ve ran this through gdb, here&#39;s the relevant output:<br>
<br>
<br>
                  #6  0x00007ffff4f6d4d5 in std::__throw_out_of_range_fmt(char const*, ...) ()<br>
                      from /usr/lib64/libstdc++.so.6<br>
                  #7  0x00000000005bb398 in _M_range_check (__n=&lt;optimized out&gt;,<br>
                       this=&lt;optimized out&gt;) at /usr/include/c++/5/bits/stl_vector.h:803<br>
                  #8  at (__n=&lt;optimized out&gt;, this=&lt;optimized out&gt;)<br>
                       at /usr/include/c++/5/bits/stl_vector.h:824<br>
                  #9  CarpetIOHDF5::AddAttributes (cctkGH=cctkGH@entry=0x1b507d0,<br>
                       fullname=fullname@entry=0x3f2434a0 &quot;ML_BSSN::cA&quot;, vdim=3,<br>
                       refinementlevel=refinementlevel@entry=0, request=request@entry=0x3df96370,<br>
                       bbox=..., dataset=83886080, is_index=false)<br>
                       at<br>
<br>
              /home/mzilhao/Trabalho/projectos/ET/Cactus/arrangements/Carpet/CarpetIOHDF5/src/Output.cc:899<br>
                  #10 0x00000000005bdea4 in CarpetIOHDF5::WriteVarChunkedParallel (<br>
                       cctkGH=cctkGH@entry=0x1b507d0, outfile=outfile@entry=16777216,<br>
                       io_bytes=@0x7fffffffc980: 1110772, request=0x3df96370,<br>
<br>
            called_from_checkpoint=called_from_checkpoint@entry=true,indexfile=indexfile@entry=-1)<br>
                       at<br>
<br>
              /home/mzilhao/Trabalho/projectos/ET/Cactus/arrangements/Carpet/CarpetIOHDF5/src/Output.cc:706<br>
                  #11 0x00000000005a233e in CarpetIOHDF5::Checkpoint (cctkGH=0x1b507d0,<br>
                       called_from=0)<br>
                       at<br>
<br>
              /home/mzilhao/Trabalho/projectos/ET/Cactus/arrangements/Carpet/CarpetIOHDF5/src/CarpetIOHDF5.cc:1277<br>
                  #12 0x000000000041f0d5 in CCTK_CallFunction (<br>
                       function=function@entry=0x5a2da0<br>
            &lt;CarpetIOHDF5::CarpetIOHDF5_InitialDataCheckpoint(cGH*)&gt;,<br>
                  fdata=fdata@entry=0x1b4a4e8, data=data@entry=0x1b507d0)<br>
                       at /home/mzilhao/Trabalho/projectos/ET/Cactus/src/main/ScheduleInterface.c:312<br>
                  #13 0x0000000000ef6499 in Carpet::CallScheduledFunction (<br>
                       time_and_mode=time_and_mode@entry=0x1174842 &quot;Meta mode&quot;,<br>
                       function=function@entry=0x5a2da0<br>
            &lt;CarpetIOHDF5::CarpetIOHDF5_InitialDataCheckpoint(cGH*)&gt;,<br>
                  attribute=attribute@entry=0x1b4a4e8, data=data@entry=0x1b507d0,<br>
                       user_timer=...)<br>
                       at<br>
<br>
              /home/mzilhao/Trabalho/projectos/ET/Cactus/arrangements/Carpet/Carpet/src/CallFunction.cc:380<br>
<br>
<br>
                  so the relevant bits of code seem to be in CarpetIOHDF5/src/Output.cc:706 and<br>
                  CarpetIOHDF5/src/Output.cc:899<br>
<br>
                  this seems to be triggered when writing hdf5 output in parallel. if i remove<br>
            checkpointing the<br>
                  run goes fine, and i do get regular 2D hdf5 output. this does not seem to be<br>
            written in<br>
                  parallel, though, as i get only one file per grid function/group. so it seems to<br>
            be the parallel<br>
                  output that triggers the crash.<br>
<br>
                  i have also tried removing all my hdf5 libs and configuring ET with<br>
            HDF5_DIR=BUILD, but the<br>
                  outcome was the same.<br>
<br>
                  has anyone seen such an error before? anything else i could provide to help<br>
            diagnose this?<br>
<br>
                  thanks,<br>
                  Miguel<br>
<br>
                  _______________________________________________<br>
                  Users mailing list<br>
            <a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a> &lt;mailto:<a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a>&gt;<br></div></div>
            &lt;mailto:<a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a> &lt;mailto:<a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a>&gt;&gt;<span class=""><br>
            <a href="http://lists.einsteintoolkit.org/mailman/listinfo/users" rel="noreferrer" target="_blank">http://lists.einsteintoolkit.org/mailman/listinfo/users</a><br>
<br>
<br>
<br>
<br>
            --<br>
            Erik Schnetter &lt;<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a> &lt;mailto:<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a>&gt;<br></span>
            &lt;mailto:<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a> &lt;mailto:<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a>&gt;&gt;&gt;<span class=""><br>
            <a href="http://www.perimeterinstitute.ca/personal/eschnetter/" rel="noreferrer" target="_blank">http://www.perimeterinstitute.ca/personal/eschnetter/</a><br>
<br>
        _______________________________________________<br>
        Users mailing list<br>
        <a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a> &lt;mailto:<a href="mailto:Users@einsteintoolkit.org" target="_blank">Users@einsteintoolkit.org</a>&gt;<br>
        <a href="http://lists.einsteintoolkit.org/mailman/listinfo/users" rel="noreferrer" target="_blank">http://lists.einsteintoolkit.org/mailman/listinfo/users</a><br>
<br>
<br>
<br>
<br>
--<br>
Erik Schnetter &lt;<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a> &lt;mailto:<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a>&gt;&gt;<br>
<a href="http://www.perimeterinstitute.ca/personal/eschnetter/" rel="noreferrer" target="_blank">http://www.perimeterinstitute.ca/personal/eschnetter/</a><br>
</span></blockquote>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">Erik Schnetter &lt;<a href="mailto:schnetter@cct.lsu.edu" target="_blank">schnetter@cct.lsu.edu</a>&gt;<br><a href="http://www.perimeterinstitute.ca/personal/eschnetter/" target="_blank">http://www.perimeterinstitute.ca/personal/eschnetter/</a></div>
</div>