CyberShake Study Database Archiving

The following procedure should be used when we want to archive a study which is either in the production or data access database to disk, to free up room in the database.

First, dump the old database contents into a new directory:

1. Determine the Study ID you want to archive, from looking at the Studies table in the DB:
select * from Studies;
2. Dump the PeakAmplitudes, using:
mysqldump --lock-all-tables --where 'Run_ID in (select R.Run_ID from CyberShake_Runs R where R.Study_ID=<study id to archive>)' CyberShake PeakAmplitudes > peak_amps.sql
3. Dump Hazard_Datasets, using:
mysqldump --lock-all-tables --where 'Study_ID=<study id to archive>' CyberShake Hazard_Datasets > hazard_datasets.sql
4. Dump Hazard_Curves, using:
mysqldump --lock-all-tables --where 'Hazard_Dataset_ID in (select D.Hazard_Dataset_ID from Hazard_Datasets D where D.Study_ID=<study id to archive>)' CyberShake Hazard_Curves > hazard_curves.sql
5. Dump Hazard_Curve_Points, using:
mysqldump --lock-all-tables --where 'Hazard_Curve_ID in (select C.Hazard_Curve_ID from Hazard_Curves C, Hazard_Datasets D where D.Study_ID=<study id to archive>)' CyberShake Hazard_Curve_Points > hazard_curve_points.sql
6. Dump CyberShake_Runs, using:
mysqldump --lock-all-tables --where 'Study_ID=<study id to archive>' CyberShake CyberShake_Runs > runs.sql
7. Dump all of the input tables -- they are small enough that we don't mind capturing data which wasn't directly used in this study.
mysqldump CyberShake AR_Hazard_Curve_Points AR_Hazard_Curves AR_Hazard_Datasets Atten_Rel_Metadata Atten_Rels CyberShake_Site_Regions CyberShake_Site_Ruptures CyberShake_Site_Types CyberShake_Sites ERF_IDs ERF_Metadata ERF_Probability_Models IM_Types Points Rupture_Variation_Probability_Modifier Rupture_Variation_Scenario_IDs Rupture_Variation_Scenario_Metadata Rupture_Variations Ruptures SGT_Variation_IDs SGT_Variation_Metadata Studies Time_Spans Velocity_Models > input_tables.sql

Next, convert the SQL dumps into SQLite format using mysql2sqlite:

1. For each of the 6 dump files, run:

 mysql2sqlite <SQL dump file> > <SQLite dump file>
 Example:  ./mysql2sqlite peak_amps.sql > peak_amps.sqlite

For large tables this may take an hour or two.

Create a SQLite database and import the tables. Use sqlite3 3.7.11 or later, or there will be an error reading the dump files.

1. Create a database for this study:
sqlite3 <study name>
2. For each dump file, run the following command:
.read <path/to/dump/file.sqlite>

Run a few queries on the original table on the database, and on the sqlite files to check that the count is the same.

1. Check the number of PeakAmplitudes:
select count(*) from PeakAmplitudes P, CyberShake_Runs R where R.Study_ID=<study_id> and P.Run_ID=R.Run_ID;
2. Check the number of rupture variations for each site:
select count(*) from CyberShake_Runs R, CyberShake_Site_Ruptures SR, Rupture_Variations V where R.Site_ID=SR.CS_Site_ID and SR.Source_ID=V.Source_ID and SR.Rupture_ID=V.Rupture_ID and R.ERF_ID=V.ERF_ID and R.ERF_ID=SR.ERF_ID and V.Rup_Var_Scenario_ID=R.Rup_Var_Scenario_ID and R.Study_ID=<study id>;

Move the files to the archive location on project.

1. Tar up the sqlite files:
tar czvf <study_name>.tgz *.sqlite
2. SFTP the tgz file to the study sqlite archive location on project at CARC (/project/scec_608/cybershake/results/sqlite_studies/<study name>), using hpc-transfer1.
3. Extract the database to /project/scec_608/cybershake/results/sqlite_studies/<study_name> .
tar xzvf sqlite_dumps/<study_name>.tgz <study_name>.sqlite