README -- GrADS Data Server v2.0 ========================================= Administrator/Maintainer -- Jennifer M. Adams (jma@cola.iges.org) Original Author -- Joe Wielgosz See COPYRIGHT for copyright information. Where to find documentation =========================== Documentation on how to install, administer and use the GDS is now in HTML. It is located in the doc/ subdirectory. You can also access the latest versions of the software and documentation on the GDS home page, http://www.iges.org/grads/gds. Release history =============== 2.0 - GDS upgraded for compatibiity with GrADS version 2.0 Requires GrADS 2.0.a2+ (executable names are different: grads and gradsdap) Handles data in grib2 format. Supports 5-Dimensional Ensemble data sets. Ensemble metadata is tailored for GrADS client, but should not interfere with access by other clients Server-side analysis is more flexible, result can vary in all 5 dimensions. Bug fix: native attributes not passed through for hdf/netcdf descriptor files 1.3 - ** GDS and Anagram are now under the GNU General Public License ** The GrADS-DODS Server was renamed to GrADS Data Server and all visible mentions of DODS replaced with OPeNDAP. Updated Java DAP to version 1.1.7 which implements DAP 2.0. Updated THREDDS XML to version 1.0. Overhaul of attribute parsing code: * descriptor attributes are distinguished from native file attributes * time metadata added for station datasets * additional metadata for gridded datasets: dataType "Grid" plus "resolution" and "resolution_units" for all grid coordinates * handles any special characters appearing in attribute values * allows overriding of data set attributes at configuration level * provides grads_* attributes sufficient for GrADS open command Improved organization of environment variables and code relating to disk paths for various server resources. The GDS shell environment is set up with bin/gds-env.sh. Multiple servers may be run from a single souce code installation by using the ANAGRAM_BASE and CATALINA_BASE environment variables to isolate the modifiable data (the config, log, and temp files) for each server instance. Added JETTY, an experimental optional servlet container. JETTY is invoked with bin/jetty-wrapper.sh and controlled by environmental variables JETTY_PORT, JETTY_STOP_PORT, and JETTY_CONFIG. 1.2.9 - Prototype support for THREDDS XML catalogs (accessed by the URL "http://servername/dods/thredds"). Contains a bug fix in subset.gs for high-res 1D subsets. This release contains an important bug fix for returning the results of analysis expressions. Results are read from the cache using the direct subsetting option, and now the data file is opened instead of the descriptor file. 1.2.8 - This release contains some minor modifications that support serving large binary files (>2 gigabytes), and high-resolution gridded data sets (>0.05 degrees lat/lon or smaller). Subsets of high-res grids will be the correct size, but server-side analysis results may not. (This is because subsets are done in grid coordinates and analysis tasks are done in world coordinates.) Further improvements to the handling of high-res grids will be a part of later GrADS releases. These GDS modifications are complemented by GrADS version 1.9b2 (and higher). This version of GrADS must be used behind the server in order for the modifications to take effect. 1.2.7 - New direct_subset attribute for dataset tag, provides an optimization for use in data-intensive applications such as distributed modeling. GDS can do in-process I/O (no separate GrADS process is invoked) for correctly-formatted IEEE binary datasets. This appears to provide close to a full order-of-magnitude improvement in throughput. Metadata attributes can be added to datasets in the configuration file using the metadata tag. Metadata attributes can be extracted from self-describing datasets using the metadata-filter tag. (Requires GrADS 1.9) A workaround for an I/O problem related to high-resolution grids has been applied. GDS can now serve datasets down to 5km resolution without distortion. Startup scripts now check the value of the GADDIR environment variable, to ensure that the sdfopen netCDF/HDF interface will function properly. Admin service can be enabled for specific privilege levels only, using the admin_allowed attribute. Import module will handle GDS 1.9 BUFR descriptors correctly. 1.2.6 - GDS now comes with Tomcat 4.1 instead of Tomcat 3.2. New start/stop/cleanup scripts in bin/: Automatically respawn Tomcat if it goes down; Can be run from any directory; Provide more informative output. New utility, utils/gds-reload.sh, provides a simple way to send admin commands to server Server-side analysis requests can now vary in three dimensions. Improved admin service design can no longer block other requests or hang server, and supports POST method for requests. Fixed bugs relating to: enabling/disabling filters abuse blocking filter not reconfiguring properly spuriously high hit counts in abuse blocking filter use of cached analysis results in follow-on analysis HTTP caching headers potential name collisions in temporary cache strided (subsampled) data requests coming up short 1.2.5 - Changed numbering scheme to remove "b". Documentation attribute seemed to confuse client parsers; changed type from URL to String. Fixed analysis bug where valid GrADS dates were being rejected Fixed data import to remove prefix and suffix strings from dataset names 1.2b4 - Added time metadata for station datasets Fixed more station data bugs 1.2b3 - Removed config file timestamp check on reconfigure Fixed directory name parsing bugs 1.2b2 - Fixed station data support 1.2b1 - Major update to GDS, with a new modular architecture, XML-based configuration mechanism, expanded documentation, major internal code improvements, and new features in the following areas: Dataset loading: - online catalog organizes datasets into subdirectories - entire directory trees and text lists of datasets can be loaded via a single entry in the config file - additional metadata attributes, and a URL pointing to online documentation, can be specified for each dataset - capable of serving remote DODS datasets, as well as datasets local to the server - if given a complete GrADS distribution, automatically invokes the correct GrADS binary (gradsc, gradsnc, gradshdf, gradsdods) for each dataset User/client interface: - more usable and consistent web page format, with some customizable hyper-links - full support for the "stride" parameter in constraint expressions - dependency checking forces recalculation of analysis results when source datasets change Basic operation: - new "admin" service can be used to control dynamic reconfiguration, and flush temporary datasets - all GDS settings are now updated during reconfiguration - multiple GDS instances can run in the same servlet container Security: - multiple IP ranges can be mapped to a single privilege set (previously known as an access level) - privilege sets can inherit settings from each other Logging: - options to send log output to a rotating set of files, or to the console - customizable logging levels and message format 1.1b9: Changed units of generate_limit setting from MB to KB for finer control. Fixed bug that caused time value at boundary of daylight savings to be off by one hour. 1.1b8: Fixed deadlock problem in dataset lock mechanism 1.1b7: Added temp_dataset_limit and temp_expire_time settings to improve control over resource usage 1.1b6: Added hit_limit and abuse_timeout settings to mitigate the effects of misbehaving clients 1.1b5: Fixed problems with using "+" in expressions, and unpredictable hangs during startup on some platforms 1.1b4: Added support for the .info and .ascii services. Made the error pages more user-friendly. Additional performance improvements. Changed missing_value to _FillValue to comply with new CF metadata convention. Log now includes user domain names when possible. 1.1b3: Added dynamic reloading of datasets, made major performance improvements. 1.1b2: Added multi-stage analysis capability, made cache persistent between restarts, and added IP security features. Fixed some minor bugs. 1.1b1: Added analysis capability, cache management, XML output, improved configuration, switched from JSDK2.1 to Tomcat 3.1, and innumerable other improvements. 1.0: First release of the GrADS Data server, with subsetting capability only. About this file =============== Last modified: $Date: 2008/07/22 17:40:55 $ Revision for this file: $Revision: 1.17 $ Release name: $Name: $ Original for this file: $Source: /homes/cvsroot/gds/gds/README,v $