Installing McIDAS-XCD v. 2018.1
Note: You must install McIDAS-X 2018.1 on the workstation before you install McIDAS-XCD 2018.1. Then continue with the appropriate set of instructions, as described below.
If you are installing McIDAS-XCD for the first time on the workstation, go to the McIDAS-XCD Administrator’s Guide‘s Installing McIDAS-XCD 2018.1 for the First Time and complete its instructions. (Use those instructions rather than those below.)
If you are installing McIDAS-XCD Fastrack version 2018.1 on a workstation that already has a previous version of McIDAS-XCD installed, complete the steps below.
- At a Unix prompt as user oper, stop McIDAS-XCD and then LDM by running the commands below in order.
Type: xcdadmin stop
Type: ldmadmin stop
- As user oper, run the script below from the ~oper/mcidas directory to build and install McIDAS-XCD 2018.1 using the same settings for optional flags -noopt and -mysql as were used to build McIDAS-X 2018.1. You must use the same compilers and optional flag settings for both the McIDAS-X and McIDAS-XCD builds.
Type: sh ./xcd2018.1.sh flags build
where flags is zero, one or both of the two optional flags described below.
- The -noopt flag is used to build McIDAS-XCD with no optimization. If you don’t specify this -noopt flag, McIDAS-XCD is built using the default optimization: -O1 for builds with gcc/gfortran.
- The -mysql flag is used to build McIDAS-XCD using the MySQL/MariaDB database libraries. Use this flag if this workstation will be serving real-time GRIB files from the NOAAPORT data stream and you have installed and want to use a MySQL/MariaDB database to make the GRIB data access faster for your McIDAS-X clients/users. If the database is in a non-standard install location, specify the flag as -mysql=directory, where directory is the root MySQL/MariaDB install directory. Remember, the -mysql flag must be specified when building both McIDAS-X and McIDAS-XCD.
When the script xcd2018.1.sh is finished, you will see the message below.
McIDAS-XCD binaries built correctly
- As user oper, run the script below from the ~oper/mcidas directory to install the new executable code and a subset of the replaceable McIDAS-XCD data files for your workstation.
Type: sh ./xcd2018.1.sh cutover
When the script is finished, you will see the message below.
McIDAS-XCD package cutover is now complete
- As user oper run the script below to make a backup copy of the ~oper/mcidas/data/ISFCDEC.CFG, ~oper/mcidas/data/SYNDEC.CFG, and ~oper/mcidas/data/NOGRIB2.CFG files then replace them with the new version of these files. The backup copies of the files are named ISFCDEC.CFG_yyyymmdd, SYNDEC.CFG_yyyymmdd and NOGRIB2.CFG_yyyymmdd, where yyyymmdd is the date when the file was backed up, e.g., 20180530.
Type: sh ./xcd2018.1.sh install 2018.1_update
- The new version of NOGRIB2.CFG by default now has three models that currently will not be decoded by McIDAS-XCD due to a large number of data points that currently cannot be displayed in McIDAS-X. They are ETSS-UNKNLC, PETS-UNKNLC, PSS-UNKNLC.
- In the new version of ISFCDEC.CFG the NCOLS variable was increased from 7500 to 10000. This means that SFCHOURLY decoder can write as many as 10,000 columns in the SFCHOURLY-schema MD files.
- In the new version of SYNDEC.CFG the NCOLS variable was increased from 6500 to 15000. This means that SYNOPTIC decoder can write as many as 15,000 columns in the SYNOPTIC-schema MD files. In addition, SYNDEC.CFG now has NROWS=24 to allow for hourly synoptic data.
- These changes, along with the STNDB.CORE updates in step 5 will result in MD file column headers being different in the MD files that are created after the STNDB.CORE updates are installed and McIDAS-XCD is restarted (last step below). This is not a problem for commands like PTLIST and PTDISP (i.e., they will continue to work with both the old and new MD files). However PTCOPY uses the column header key values from the first MD file in the dataset when making its output MD file. So PTCOPYs of MD files created after the restart will have the old/incorrect column header keys in their output files until the first MD file in the dataset is replaced (which could take as long as nine days in a real-time cylinder).To avoid this problem you can delete all the real-time MD files prior to restarting McIDAS-XCD (last step below). However that would result in the loss of all real-time MD file data from prior days, which may not be necessary if your users are mainly using commands like PTLIST, PTDISP, SFCPLOT, RABOCON, etc. instead of PTCOPY. Each site will need to decide on its own how they want to address it. If you decide to delete all the real-time MD file, you can optionally do so by running command BATCH RMMDFILES.BAT in a McIDAS-X session started as user oper.
- As user oper, start a McIDAS-X session and run the command below to remove the .IDT files. The .IDT files are recreated after McIDAS-XCD is restarted (last step below) and at that point will contain the STNDB.CORE updates.
Type: BATCH RECREATEIDT.BAT
- In the McIDAS-X session started as user oper in step 5 above, run the command below to update the SCHEMA files on the McIDAS-XCD machine. This is necessary due to the updates to a number of SYNOPTIC, SFCHOURLY and RAOB stations as well as the transition to hourly SYNOPTIC data.
Type: BATCH UPDATESCHEMA.BAT
- In the McIDAS-X session started as user oper in step 5 above, run the two commands below. The first command (BATCH REGROUP.BAT) recreates the GROUPS.DAT and COUNTRY.DAT files so they contain the station database changes included in McIDAS-X Fastrack 2018.1. The second command (BATCH REBILDRAP.BAT) rebuilds the Rapid Access Pointer files so they include the same station database changes and rebuilds the Rapid Access Text files. Note that running the second command erases the previous content of the *.RAP and *.RAT files, so they will contain only new data. That means that the user commands MOSRPT, OBSRPT, RAOBRPT, SFCRPT, SYNRPT and TAFRPT will not be able to access any data from prior to McIDAS-XCD being restarted (last step below).
Type: BATCH REGROUP.BAT
Type: BATCH REBILDRAP.BAT
- As user mcadde, start a McIDAS-X session and run the command below to recreate the GROUPS.DAT and COUNTRY.DAT files in the mcadde account so they contain the station database changes included in McIDAS-X Fastrack 2018.1.
Type: BATCH REGROUP.BAT
- As user mcadde, make a backup copy of your RESOLV.SRV file (e.g., cp RESOLV.SRV RESOLV.SRV.BAK) in preparation of adding the restructured RTGRIDS/* datasets in steps 10 and 11 below.
- In the McIDAS-X session started as user mcadde in step 8 above, run the command below to remove the obsolete and/or redefined RTGRIDS/* datasets.
Type: BATCH RMXCDGRIB2018_1.BAT CONTINUE=YES
- In the McIDAS-X session started as user mcadde in step 8 above, run the command below to update the RTGRIDS/* datasets so they contain the data currently available over the NOAAPORT WMO and NGRID feeds.
Type: BATCH XCDNOAAPORTGRID.BAT
- The XCDADDE.BAT file, which is optionally run to create the default ADDE datasets, includes XCDNOAAPORTGRID.BAT among the batch files it runs. See Configuring the Default ADDE Datasets in the McIDAS-XCD Administrator’s Guide for more information. As also noted there, you can optionally run the command BATCH XCDCONDUITGRID.BAT to create ADDE datasets for the subset of CONDUIT LDM GRIB2 grids that the SSEC Data Center provides for its users.
- The XCDNOAAPORTGRID.BAT and XCDCONDUITGRID.BAT files were first implemented in McIDAS-XCD 2014.1; see the 2014.1 software and instructions page for more information.
- As user mcidas, run the script below from the ~oper/mcidas directory to install the new ADDE server executable code in the mcidas account.
Type: sh ./xcd2018.1.sh install addeservers
- Restart LDM and then McIDAS-XCD by running the commands below in order at a Unix prompt as user oper.
Type: ldmadmin start
Type: xcdadmin start
This concludes the McIDAS-XCD Fastrack version 2018.1 installation procedure. If you would like to modify the configuration of your McIDAS-XCD workstation, go to the McIDAS-XCD Administrator’s Guide‘s Configuring McIDAS-XCD.