<html><head></head><body>Yesterday I had the same issue with a CI run of dune-istl. It seems this is related to MPI.<br><br>Ciao<br>Christian<br><br><div class="gmail_quote">Am 28. September 2020 08:55:58 MESZ schrieb Oliver Sander <oliver.sander@tu-dresden.de>:<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<pre class="k9mail">I just noticed the following in<br><br> <a href="https://gitlab.dune-project.org/kilian.weishaupt/dune-grid/-/jobs/175885">https://gitlab.dune-project.org/kilian.weishaupt/dune-grid/-/jobs/175885</a><br><br><br>[1601275211.221338] [runner-d307b235-project-936-concurrent-0:11175:0]       mm_posix.c:162  UCX  ERROR Not enough memory to write total of 4292720 bytes. Please check that /dev/shm or the directory you specified has more available memory.<br>3773[1601275211.221802] [runner-d307b235-project-936-concurrent-0:11175:0]        uct_mem.c:132  UCX  ERROR failed to allocate 4292720 bytes using md posix for mm_recv_desc: Out of memory<br>3774[1601275211.222012] [runner-d307b235-project-936-concurrent-0:11175:0]          mpool.c:191  UCX  ERROR Failed to allocate memory pool (name=mm_recv_desc) chunk: Out of memory<br>3775[1601275211.222397] [runner-d307b235-project-936-concurrent-0:11175:0]       mm_iface.c:644  UCX  ERROR failed to get the first receive descriptor<br>3776[runner-d307b235-project-936-concurrent-0:11175] ../../../../../../ompi/mca/pml/ucx/pml_ucx.c:291  Error: Failed to create UCP worker<br>3777[runner-d307b235-project-936-concurrent-0:11175] [[28993,1],1] selected pml ob1, but peer [[28993,1],0] on runner-d307b235-project-936-concurrent-0 selected pml ucx<br>3778--------------------------------------------------------------------------<br>3779MPI_INIT has failed because at least one MPI process is unreachable<br>3780from another.  This *usually* means that an underlying communication<br>3781plugin -- such as a BTL or an MTL -- has either not loaded or not<br>3782allowed itself to be used.  Your MPI job will now abort.<br>3783You may wish to try to narrow down the problem;<br>3784 * Check the output of ompi_info to see which BTL/MTL plugins are<br>3785   available.<br>3786 * Run your application with MPI_THREAD_SINGLE<br><br><br>@CI_gurus can you please have a look!<br><br>Thanks,<br>Oliver<br><br><br>On 27.09.20 09:10, Oliver Sander wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #729fcf; padding-left: 1ex;">Dear Dune,<br><br>the CI system for the master branch seems to fail with the debian 11 gcc-9-20 image:<br><br>  <a href="https://gitlab.dune-project.org/core/dune-grid/-/pipelines/29754">https://gitlab.dune-project.org/core/dune-grid/-/pipelines/29754</a><br><br>It reports a run-time failure in some YaspGrid-related tests.  Any idea<br>about the possible causes of this?<br><br>Best regards,<br>Oliver<hr>Dune-devel mailing list<br>Dune-devel@lists.dune-project.org<br><a href="https://lists.dune-project.org/mailman/listinfo/dune-devel">https://lists.dune-project.org/mailman/listinfo/dune-devel</a><br><br></blockquote><br></pre></blockquote></div></body></html>