-------------------------- Darshan Release Change Log -------------------------- Darshan-3.1.5-pre1 ============= * modify Darshan MPI instrumentation method to intercept both MPI and PMPI symbols to workaround MPI implementations that are calling PMPI routines directly (e.g., the Fortran bindings of OpenMPI2). Contributed in part by Chris Zimmer. * add a new python utility for analzying DXT trace files (dxt_analyzer). Contributed by Alex Sim. * bug fix to disable Darshan module instrumentation after the Darshan shutdown procedure has began, an issue that was leading to negative timers in some cases * bug fix for autoconf ignoring specified libbz2 location. Contributed by Glenn Lockwood. * add regression tests harnesses for Cray systems at ALCF & NERSC * add support for DARSHAN_EXCLUDE_DIRS environment variable to explicitly disable instrumentation for files in given directories. Contributed by Cristian Simarro. Darshan-3.1.4 ============= * bug fix to prevent darshan-parser segfault when parsing logs with DXT module data present. Reported by James Dickson. * bug fixes to make the darshan-diff utility functional again. Darshan-3.1.3 ============= * disable instrumentation for mmap when dynamically linking; this avoids a potential deadlock condition on Cray systems using dynamically linked executabes. Reported by Cristian Simarro for dynamic linking case. Darshan-3.1.3-pre1 ============= * add new DxT instrumentation modules to provide fine-grained read/write operation tracing at both the POSIX and MPI-IO layers - this functionality should be enabled at runtime by exporting the DXT_ENABLE_IO_TRACE environment variable - trace output is stored within Darshan's traditional log file format - a corresponding trace parser (darshan-dxt-parser) is offered within darshan-util to allow the DxT trace modules to be parsed and displayed - this software was contributed by Cong Xu and Intel's HPDD division. * add logic to allow Darshan to capture command line arguments from Fortran applications (contributed by Cristian Simarro) * skip instrumentation attempts for anonymous mmap() calls; this avoids a potentential deadlock condition when used with hugepages on Cray systems. Reported by Glenn Lockwood for static linking case. * fix segmentation fault in statistics collection for applications that issue operations with a large number of distince access sizes or strides on the same file. Reported by Glenn Lockwood. * disable HDF5 module by default unless enabled using --enable-HDF5-post-1.10 or --enable-HDF5-pre-1.10 configure arguments. These options vary the wrapper prototypes to match the corresponding HDF5 library ABI. The initial patch for HDF5 1.10 compatibility was contributed by Karl-Ulrich Bamberg. * modified Darshan's path exclusion logic to include a whitelist to prevent I/O to/from Cray's Datawarp service from being filtered out (it is located within the /var directory, which historically has been excluded). Reported by Glenn Lockwood. Darshan-3.1.2 ============= * bug fix in resolving underlying call to fopen64 when using the LD_PRELOAD instrumentation mechanism * bug fix to make sure Lustre module headers are installed when installing darshan-util - reported by Matthieu Dorrier Darshan-3.1.1 ============= * bug fix to MPIIO_F_WRITE_START_TIMESTAMP, which may have produced incorrect timestamps in some cases - reported by Wucherl (William) Yoo Darshan-3.1.0 ============= * add stdio I/O library instrumentation module (Philip Carns) - this handles instrumentation of file stream I/O functions like fopen(), fprintf(), fscanf(), etc. - this module also captures stats on the standard streams (stdin, stdout, & stderr) * add Lustre instrumentation module (Glenn Lockwood) - this module provides Lustre striping details (e.g., stripe width, stripe size, list of OSTs a file is striped over) * add new mmap-based logging mechanism that allows Darshan to generate output logs even in cases where applications don't call MPI_Finalize() - these logs are uncompressed and are per-process rather than per-job * add the darshan-merge utility to darshan-util to allow per-process logs generated by the mmap-based logging mechanism to be converted into Darshan's traditional compressed per-job log files * augment the POSIX module timestamp counters to also include a LAST_OPEN & FIRST_CLOSE counters to give more details on application I/O intervals * avoid saving duplicate mount point entries in Darshan log files Darshan-3.0.1 ============= * bug fix in darshan logutil mount parsing code that was causing file paths to be matched to the first mount point with a common prefix rather than the one with the longest common prefix * bug fix in the darshan-util bzip2 configure check that was accidentally overwriting Darshan's LDFLAGS * minor bug fixes to IO start time counters in all modules to set IO start time to the actual first start time rather than the first IO op to complete * update darshan-util perl scripts to get perl bin from user's path, rather than from /usr/lib (reported by Kay Thust) * update Darshan's fortran and cxx compiler wrapper generators to automatically detect MPICH library names on BG/Q * fix bug that was calculating Darshan's agg_perf_by_slowest performance metric incorrectly * add performance estimate info to darshan-job-summary graphs Darshan-3.0.0 ============= * install darshan-null-log-format.h header when installing darshan-util component, otherwise compiler errors are generated when building external tools that use libdarshan-util * update docs to give debugging tips for cases where Darshan logs are not generated * fix shared library regression test script to check for potential errors with Darshan symbols rather than failing silently in these cases * bug fix for determining minimum non-zero counters in shared file reductions in all modules * loosen Darshan's PMPI symbol check to prevent inadvertent disabling of Darshan for some MPICH builds * update runtime docs to give information on upgrading Darshan * bug fix for resolving MPI_Gather and MPI_Barrier when LDPRELOADing Darshan's shared libraries (reported by Richard Hedges and Rob Latham) * add more helpful error handling when opening 2.x version log files * port darshan-diff utility over to new log file format * fix numerous configure bugs on Cray systems * add synthetic benchmarking hooks for testing Darshan's shutdown procedure Darshan-3.0.0-pre3 ============= * add module-specific version fields to header to allow utilities to handle different versions of a module's I/O data for backwards compatibility -- NOTE: this breaks the log file parsing for logs obtained using Darshan-3.0.0-pre2 & Darshan-3.0.0-pre1 * bug fix in regression test scripts for setting proper environment variables to use MPI profiling configuration for Fortran apps * bug fix in bzip2 log writing implementation in darshan-logutils * possible race conditions resolved in each module's shutdown code * general code, comment, and documentation cleanup * addition of module-specific counter descriptions printed prior to parsing a modules I/O data in darshan-parser Darshan-3.0.0-pre2 ============= * add fix to install appropriate headers for linking external applications with darshan-util (reported by Matthieu Dorier) * add darshan-util Ruby bindings for the new modularized version of Darshan (3.0) (Matthieu Dorier) * add enhancement to darshan-runtime to allow per-module instrumentation memory to be user configurable using a configure option or a runtime environment variable Darshan-3.0.0-pre1 ============= * new version of Darshan with the following features/improvements: - hooks for developers to add their own instrumentation module implementations to capture new I/O characterization data - these instrumentation modules can be used to instrument new I/O interfaces or gather system-specific parameters, for instance - modularized log format allows new module-specific utilities to access their I/O characterization data independently - this new format also allows new counters to be added to existing instrumentation modules without breaking existing utilities - Darshan logs now contain a mapping of Darshan's unique record identifiers to full file names, instead of fix-sized file name suffixes - a new instrumentation module for capturing BG/Q-specific parameters (BG/Q environment is automatically detected at configure time) (implemented by Kevin Harms) - new darshan-parser and darshan-job-summary output to utilize the new modularized log format * updated documentation outlining changes in this release, as well as steps for adding new instrumentation modules is given in the top-level 'doc' directory. - documentation for configuring and using the darshan-runtime and darshan-util components are mostly the same and still located in their respective directories ('darshan-runtime/doc' and 'darshan-util/doc') darshan-2.3.2-pre1 ============= * Fix gnuplot version number check to allow darshan-job-summary.pl to work with gnuplot 5.0 (Kay Thust) * Fix function pointer mapping typo in lio_listio64 wrapper (Shane Snyder) * Fix faulty logic in extracting I/O data from the aio_return wrapper (Shane Snyder) * Fix bug in common access counter logic (Shane Snyder) * Expand and clarify darshan-parser documentation (Huong Luu) darshan-2.3.1 ============= * added documentation and example configuration files for using the -profile or $MPICC_PROFILE hooks to add instrumentation to MPICH-based MPI implementations without generating custom wrapper scripts * Add wrappers for mkstemp(), mkostemp(), mkstemps(), and mkostemps() (reported by Tom Peterka) * Change OPEN_TIMESTAMP field to report timestamp right before open() is invoked rather than after timestamp after open is completed. NOTE: updated log format version to 2.06 to reflect this change. * Change start_time and end_time fields in job record to use min and max (respectively) across all ranks * Fix bug in write volume data reported in file system table in darshan-job-summary.pl (reported by Matthieu Dorier) * Clean up autoconf test for zlib and make zlib mandatory (reported by Kalyana Chadalavada) * add --start-group and --end-group notation to Darshan libraries for Cray PE 2.x environment to fix link-time corner cases (Yushu Yao) * improve y axis labels on time interval graphs in darshan-job-summary.pl (reported by Tom Peterka) * misc. improvements to darshan-parser --perf output (reported by Shane Snyder) - indicate which rank was slowest in unique file results - label I/O vs. meta time more clearly - include unique file meta time in agg_perf_by_slowest calculation * added regression test script framework in darshan-test/regression/ - currently support platforms include: - Linux with static linking and generated compiler wrappers - Linux with static linking and profiler configuration files - Linux with dynamic linking and LD_PRELOAD - Blue Gene/Q with static linking and profiler configuration files * update darshan-gen-fortran.pl and darshan-gen-cxx.pl to support new library naming conventions in MPICH 3.1.1 and higher * update documentation to reflect known issues with some versions of MPICH * Cray platforms: modify darshan-runtime so that link-time instrumentation options are only used when statically linking via Libs.private. (reported by Kalyana Chadalavada) darshan-2.3.0 ============= * Fix incorrect version numbering in darshan-runtime component of Darshan 2.2.9, reported by Jean-Guillaume Piccinali darshan-2.2.9 ============= * Bug fixes: - Fix mnt table overflow if a large number of file systems are mounted, reported by David Shrader. - Fix argument parsing for darshan-convert, reported by Mouhamed Gueye. - Fix metadata annotation overflow in darshan-convert, reported by Mouhamed Gueye. - Fix const-correctness in dynamic library when built against MPI 3.x libraries. - Fix "undefined symbol: dlsym" error when using preloaded dynamic library on some platforms, reported by Florin Isaila. - Normalize timestamps to always be relative to MPI_Init(). - Better library name matching in compiler wrappers to handle more MPICH variations on Blue Gene systems. - Fixed header dependency problems when using darshan-logutils.h in external programs (Shane Snyder). * Enhancements: - Add support (both in documentation and in provided module files) for Cray PE 2.x. - Honor CC variable to allow darshan-util to be built with other compilers besides gcc. - Update documentation to clarify the interaction between rpath and LD_PRELOAD when instrumenting dynamic libraries, issue reported and investigated by Davide Del Vento. - Ability to disable shared-file reduction by setting the DARSHAN_DISABLE_SHARED_REDUCTION environment variable. - More thorough output from darshan-parser --perf, suggested by Huong Luu. - Increased metadata annotation room from 64 bytes to 1KiB in header. - CP_F_{FASTEST/SLOWEST}_RANK_TIME counters now take MPI-IO time into account, not just POSIX time, issue reported by Huong Luu. - Better handling of systems with many mounted file systems (after which point Darshan will assume file resides on / file system), issue reported and investigated by David Shrader: - Track up to 64 rather than 32 mounted file systems at runtime. - Increase header space available for storing mount point information in log file from approximately 1 KiB to approximately 3 KiB. - Prioritize storing information about non-NFS volumes over NFS volumes if too many file systems are mounted to record them all. - Added darshan-util pkgconfig file (Shane Snyder). - Added --enable-shared configure option to darshan-util to build and a shared library version of libdarshan-util darshan-2.2.8 ============= * WARNING: please note that the Darshan module file for Cray environments has been updated, especially in the DARSHAN_POST_LINK_OPTS variable. Please update your module file accordingly when upgrading from 2.2.7 or earlier on Cray platforms. * Improved ability to analyze I/O activity related to particular files opened by an application - darshan-summary-per-file.sh script to generate a separate pdf summary for each file opened by an application. Developed by Rob Latham. - Added --file-list and --file-list-detailed options to darshan-parser to list files opened by an application along with brief statistics - Added --file option to darshan-convert to filter out activity for a specific file from a Darshan log * Add wrappers for POSIX AIO operations (fixes tracking of underlying POSIX operations resulting from nonblocking MPI-IO operations. Bug reported by David Shrader.) * Fix compile-time errors when Darshan is configured for use with MPICH 3.x installations with const correct MPI function prototypes. Also affects MPICH 1.5+ installations with optional const support. Reported by Yushu Yao. * Fix segmentation fault when using LD_PRELOAD instrumentation on programs that use MPI_Init_thread() rather than MPI_Init(). Reported by Myriam Botalla. darshan-2.2.7 ============= * Updated Cray installation documentation for cleaner integration in Cray environments * Fix bug that recorded incorrect device ID (and therefore incorrect mount point mapping) if stat() was called before open() on a file * Store version number of the darshan runtime library in the log file metadata (see lib_ver in darshan-parser output) darshan-2.2.6 ============= * Bug fixes: - make sure to honor user-specified hints passed in at runtime via the DARSHAN_LOGHINTS env variable. - include fread and fwrite in darshan-job-summary.pl read and write counts - fix segmentation fault on invalid arguments to darshan-parser * Performance enhancements: - collect mount point information at rank 0 and broadcast to all processes to avoid excess file system traffic on startup - change default MPI-IO hints for writing log file to romio_no_indep_rw=true and cb_nodes=4 to improve log creation performance * Install libdarshan-util and headers during installation process for darshan-util; patch contributed by Matthieu Dorier * Detect PMPI support at link time when using compiler scripts produced by the darshan-gen-* utilities. This avoids link problems when Darshan compiler wrappers are used with the ADIOS dummy MPI library; reported by Jingqing Mu * Support MPICH_{CC/CXX/F77} environment variables in compiler scripts produced by the darshan-gen-* utilities * Add instrumentation of startup time, enabled via DARSHAN_INTERNAL_TIMING environment variable * Rename cp-shutdown-bench test utility to darshan-shutdown-bench and enable benchmarking hooks in library by default so that darshan-shutdown-bench can be used with any Darshan installation * Remove deprecated --enable-st-dev-workaround configure option * Misc. documentation updates darshan-2.2.5 ============= * Fix bug in mount point identification when --enable-stat-at-open option is not used. In Darshan 2.2.4, some file entries were recorded as using the "/" file system regardless of their location. * Update patches and documentation for Cray xt-asyncpe environment 5.12 or higher; contributed by Yushu Yao. * Update documentation to no longer recommend --enable-stat-at-open option on any systems * Fix unresolved symbols error when linking non-MPI programs with Cray compiler. Reported by Yushu Yao. * Use fstat() in place of stat() for detecting file information when --enable-stat-at-open option is used * Use statfs() at startup to detect default block size for supported file systems * Added --enable-group-readable-logs configure option, which will cause Darshan to to generate log files with the group read permission bit set. This option is useful in conjunction with deployments that set the setgid bit on log directories. darshan-2.2.4 ============= * Disable extra stat() of newly opened files by default. This improves performance on shared files for some platforms. Reported by Yushu Yao. - prevents Darshan from reporting SIZE_AT_OPEN, FILE_ALIGNENT, and DARSHAN_FILE_NOT_ALIGNED values - features can still be enabled via --enable-stat-at-open option * Fix missing -lz in post ld flags reported by Yushu Yao. Fixes a link-time error for some corner-case applications. * Fix bug in Cray compiler script patches that was setting compiler flags incorrectly. Reported by Yushu Yao. WARNING: if you are using a previous Darshan release (2.2.3 through 2.2.4-pre4) on a Cray platform, please re-patch your compiler scripts. * Update darshan-gen-* scripts to support the potential for additonal LDFLAGS or CFLAGS that appear before the object or source file in the compile or link commands. This fixes compatibility with some mvapich2 installations, reported by Dragos Constantin. darshan-2.2.3 ============= * improved Cray XE6 support - support for GNU, PGI, Cray, Pathscale, and Intel compilers - patch adding Darshan capability to system compiler scripts - software module, including testing and features contributed by Yushu Yao and Katie Antypas of The National Energy Research Scientific Computing Center (NERSC) - improved documentation * minor bug fixes: - properly detect cxx library name when generating BG/Q compiler wrappers - improve timing mechanism for cp-shutdown-bench benchmark - improve hashing to avoid log file name collisions darshan-2.2.2 ============= * bug fixes: - remove debugging message that was inadvertently included in MPI_File_sync() wrapper - fix potential hang if the --with-log-path-by-env argument was used at configure time but the environment variable was not set at run time darshan-2.2.1 ============= * significant improvements to how counters are handled in multi-threaded applications * initial (rough) documentation for using Darshan in Cray programming environments with static linking * bug fixes: - escape special characters in mount point paths in darshan-job-summary.pl (reported by Mouhamed Gueye) - workarounds for various runtime problems with cuserid() and stat() in Cray environments - build problems with darshan-utils on some versions of OSX - accurate shared file statistics for libraries that use deferred opens darshan-2.2.0 ============= * split darshan into separate packages: - darshan-runtime: for runtime instrumentation - darshan-utils: for processing darshan log files * changed default output file name for darshan-job-summary.pl to be based on input file name rather than summary.pdf * reorganized init and finalize routines so that can be linked separately (to allow for easier integration with other instrumentation tools) * add -cc, -cxx, -f77, -f90, and -fc support to compiler scripts generated by the darshan-gen-*.pl scripts * bug fixes: - potential MAX_BYTE overflow on 32 bit systems - incorrect pread and pwrite offset tracking - corrections to darshan-job-summary variance table - better runtime error handling if bzip or gnuplot tools are insufficient - improvements to time range in darshan-job-summary graphs * documentation: - improved documentation for both the darshan-runtime and darshan-util portions of Darshan can be found in the respective doc/ subdirectory for each darshan-2.1.2 ============= * improved error handling when writing log files. If a write fails then the log file will be deleted and a warning will be printed to stderr. darshan-2.1.1 ============= * new darshan-convert command line utility for converting existing log files, with optional anonymization and optional bzip2 compression * bzip2 support in command line utilities (but not in the darshan library itself) * updated log file format that allows for string key/value pairs to be stored in the header * added ability to set MPI-IO hints when writing darshan log at configure time: --with-log-hints at run time: DARSHAN_LOGHINTS environment variable * bug fix contributed by Sandra Schröder: use case-insensitive search for MPI symbols in Fortran wrapper script * performance bug fix: remove unecessary call to MPI_File_set_size when writing log * added --with-logpath-by-env configure option to allow absolute log path to be specified via environment variable darshan-2.1.0 ============= * additinoal environment variables to control log, jobid and alignment parameters * additional configure tests to improve portability * bug fixes for darshan-parser --perf calucations * support for MPI1.x * support for OpenMPI * support for PGI, Intel compilers * new libdarshan.so dynamic library for use with LD_PRELOAD darshan-2.0.2 ============= * added a random identifier to job logs (to avoid collisions from multiple application instances within a single scheduler job) * improved installation and library path management for darshan-job-summary.pl * improved error handling in darshan-job-summary.pl * additional derived statistics categories for darshan-parser output: --all : all sub-options are enabled --base : darshan log field data [default] --file : total file counts --perf : derived perf data --total : aggregated darshan field data darshan-2.0.1 ============= * bug fix to variance/minimum calculations on shared files * switch to automatic generation of all MPI compiler scripts using darshan-gen-* tools * new run time environment variable: DARSHAN_INTERNAL_TIMING. If set at job execution time, it will cause Darshan to time its own internal data aggregation routines and print the results to stdout at rank 0. darshan-2.0.0 ============= * new output file format that is portable across architectures NOTE: Darshan 1.x output files are incompatible with the tools in this release unless they were generated on a ppc32 architecture (Blue Gene) * 8 new counters that record the rank of the fastest and slowest process that opened each shared file, along with the number of seconds and number of bytes consumed by those processes. It also reports the variance in both time and amount of data. * new --with-jobid-env configure argument to support recording job identifiers from different schedulers * job ID is now recorded within the Darshan log in addition to in the file name * darshan-job-summary.pl: * opens output files directly without using intermediate darshan-parser output * table showing data usage per file system * table showing I/O variance in shared files * Fixes for bugs reported by Noah Watkins: * avoid name collision in hashing function * divide by zero error in darshan-job-summary.pl darshan-1.1.14 ============== * fixed erroneous incompatibility warning when opening old logs in darshan-parser darshan-1.1.13 ============== * updated compiler scripts to fix link order in some cases with -lhdf5 darshan-1.1.12 ============== * improved mapping of file records to mount points * new page in darshan-job-summary.pl output showing timelines of file access * bug fix to "fast" mpif90 script * checkpoint/restart ability in parallel fsstats script darshan-1.1.11 ============== * Bug fix for lseek, pread, and pwrite when used in 32 applications without large file support * Improved experimental darshan-gen-cc.pl script * Added new experimental darshan-gen-cxx.pl script darshan-1.1.10 ============== * Bug fix for files that are accessed with stat() but never opened * Workaround zlib problems with 64 bit offsets on 32 bit architectures darshan-1.1.9 ============= * Added "fast" version of each BG/P compiler * Added experimental scripts in test directory to run fsstats in parallel * Kevin Harms: Added experimental utilities for loading darshan results into SQL * Use exclusive flag when opening output file (to protect against file name collision) * Rob Ross: updates to allow command line tools to build on Darwin * Bug fix for pnetcdf configure problem reported by Rob Latham; darshan now always pulls in MPI_Wtime() symbol at link time darshan-1.1.8 ============= * Added tracking of file system type and mount point for each file * Added tracking of file size at open time (CP_SIZE_AT_OPEN) * Moved sync cost to be counted in cumulative write time rather than cumulative metadata time * Added sync as a separate category in darshan-job-summar.pl counters * Bug fix to most frequent access size table in darshan-job-summary.pl * Converted all utilities to use darshan-logutils api for reading output files * Added backwards compatibility to darshan-logutils routines * Kevin Harms: Added darshan-analyzer utility to summarize usage of MPI-IO, pNetCDF, HDF5, and shared files across a set of output files * Fixed bug field listing for darshan-diff utility darshan-1.1.7 ============= * Minor fix for a compile warning darshan-1.1.6 ============= * Kevin Harms: bug fix for segfault in apps that use MPI_Init_thread() darshan-1.1.5 ============= * Limit PMPI usage in library to fewer functions * Update PMPI detection in compiler scripts to ignore functions unused by Darshan darshan-1.1.4 ============= * Track files opened via Parallel NetCDF * Track files opened via HDF5 * Record slowest individual POSIX read and write times along with access size for those operations * Inspect symbols at compile time to determine whether to enable Darshan or not based on the presence of MPI and PMPI symbols * Use GNU and IBM compilers from path rather than hard coded location * Simplify warning message if unable to open log file * Remove unused internal benchmark routines darshan-1.1.3 ============= * Updated compiler scripts for V1R4 driver on BlueGene/P darshan-1.1.2 ============= * Added *_r versions of each IBM compiler script on BlueGene/P darshan-1.1.1 ============= * Set default permissions to 0400 (user read only) for output files * Automatically disable Darshan at link time if common PMPI libraries are detected in the command line * Experimental tool (darshan-gen-cc.pl) to automatically generate Darshan-enabled mpicc scripts darshan-1.1.0 ============= * Initial public release