>From bcc8c1aa5672331b10671a3ae458ac17bea67a91 Mon Sep 17 00:00:00 2001 From: Rob Latham Date: Mon, 24 Mar 2014 16:49:18 -0500 Subject: [PATCH] fix fs detection when multiple fs exist ROMIO code assumes all processes will use the same ROMIO driver. we were not reaching the "find a common file system" logic when NFS was enabled, everyone stat-ed the file system without errors, but some processees found a different file system (like if some processes are writing to NFS and others to UFS) See discussion beginning here: http://lists.mpich.org/pipermail/discuss/2014-March/002403.html --- src/mpi/romio/adio/common/ad_fstype.c | 42 ++++++++++++++++------------------- 1 file changed, 19 insertions(+), 23 deletions(-) diff --git a/src/mpi/romio/adio/common/ad_fstype.c b/src/mpi/romio/adio/common/ad_fstype.c index ca971291..365ce480 100644 --- a/src/mpi/romio/adio/common/ad_fstype.c +++ b/src/mpi/romio/adio/common/ad_fstype.c @@ -686,32 +686,28 @@ void ADIO_ResolveFileType(MPI_Comm comm, const char *filename, int *fstype, } } else { ADIO_FileSysType_fncall(filename, &file_system, &myerrcode); - if (myerrcode != MPI_SUCCESS) { - *error_code = myerrcode; - /* the check for file system type will hang if any process got - * an error in ADIO_FileSysType_fncall. Processes encountering - * an error will return early, before the collective file - * system type check below. This case could happen if a full - * path exists on one node but not on others, and no prefix - * like ufs: was provided. see discussion at - * http://www.mcs.anl.gov/web-mail-archive/lists/mpich-discuss/2007/08/msg00042.html - * (edit: now - * http://lists.mcs.anl.gov/pipermail/mpich-discuss/2007-August/002648.html) - */ - - MPI_Allreduce(error_code, &max_code, 1, MPI_INT, MPI_MAX, comm); - if (max_code != MPI_SUCCESS) { - *error_code = max_code; - return; - } - /* ensure everyone came up with the same file system type */ - MPI_Allreduce(&file_system, &min_code, 1, MPI_INT, - MPI_MIN, comm); - if (min_code == ADIO_NFS) file_system = ADIO_NFS; + /* the check for file system type will hang if any process got + * an error in ADIO_FileSysType_fncall. Processes encountering + * an error will return early, before the collective file + * system type check below. This case could happen if a full + * path exists on one node but not on others, and no prefix + * like ufs: was provided. see discussion at + * http://www.mcs.anl.gov/web-mail-archive/lists/mpich-discuss/2007/08/msg00042.html + * (edit: now + * http://lists.mcs.anl.gov/pipermail/mpich-discuss/2007-August/002648.html) + */ + + MPI_Allreduce(&myerrcode, &max_code, 1, MPI_INT, MPI_MAX, comm); + if (max_code != MPI_SUCCESS) { + *error_code = max_code; + return; } + /* ensure everyone came up with the same file system type */ + MPI_Allreduce(&file_system, &min_code, 1, MPI_INT, + MPI_MIN, comm); + if (min_code == ADIO_NFS) file_system = ADIO_NFS; } - } else { /* prefix specified; just match via prefix and assume everyone got -- 1.8.3.2