Import CSV files with tsload

To add data into the system, you must have the required permissions. Initially, only the Admin user has that level of access.

Use ThoughtSpot Loader (tsload) to load data from a CSV text file into an existing table in ThoughtSpot. ThoughtSpot Loader (tsload) is a common way to import data. When using tsload, you can load larger datasets and make the loading process repeatable through scripting.

The tsload command accepts flags that enable you to specify column and row separators, date or timestamp formats, null value representations, and similar parameters. Many of these options have defaults that you can override.

Before importing data, you need to build the schema.

After importing data, you can see it in the Data tab.

ThoughtSpot Loader

To use ThoughtSpot Loader, type the command tsload followed by the appropriate flags. You can see the list of the flags it accepts in the ThoughtSpot Loader flag reference, or by issuing tsload -help on the command line.

The tsload feature supports both full and incremental data loads. For incremental loads, it performs an upsert (insert or update). If an incoming row has the same primary key as an existing row, it updates the existing row with new values.

You can integrate tsload into your ETL environment for more automated data loads. Most ETL tools enable you to write target data into files, and support scripted post-transformation actions that include loading data into ThoughtSpot.

Load data using tsload

To manually load data into ThoughtSpot, follow these steps.

You can also save the tsload commands to a script, for reuse.

  1. Log in to the Linux shell using SSH.

    $ ssh admin@<cluster-IP>
  2. Change to the directory that contains the staged CSV files.

    $ cd <directory>
  3. Use the following syntax to invoke tsload, specifying the appropriate flags, and your data source file:

    $ tsload --target_database=<my_database>
           --target_table=<my_table> --alsologtostderr
           --empty_target --source_file=<my_file.csv> --v 1
           --field_separator="separator_char"

    This example imports the CSV file ssbm_customer.csv into the table CUSTOMER:

    $ tsload --target_database=SAMPLE_DB
           --target_table=CUSTOMER --alsologtostderr
           --empty_target --source_file=ssbm_customer.csv
           --v 1 --field_separator "|"
  4. When tsload starts processing the data, you can see messages that indicate the progress. When the load is complete, you can see two summary messages: source summary and load summary.

     Started processing data row
     Source has 32 data rows, ignored row count 0
     Waiting for rows to commit...(please wait)
    
     Source summary
     --------------
     Data source:                     ssbm_customer.csv
     Source data format:              csv
     Header row?:                     no
     Tokenizer Options:               escape_char: "" field_separator: "|" enclosing_char: "\""
     Date format:                     %Y%m%d
     Date time format:                %Y%m%d %H:%M:%S
     Flexible mode?:                  no
    
     Load summary
     ------------
     Target table:                    CUSTOMER
     Should empty target?:            yes
     Status:                          Successful
     Rows total:                      32
     Rows successfully loaded:        30
     Rows failed to load:             0
     Rows duplicate/omitted:          2
     % of Rows successfully loaded:   93.75 %
     Load Rate (MB/s):                0.00 MB/s
     Load Rate (Rows/s):              3.53 Rows/s
     Start time (Wallclock):          Tue Jan 29 09:09:07
     End time (Wallclock):            Tue Jan 29 09:09:08
     Total load time = 1.13 seconds = 0.02 minutes = 0.00 hours
     Data size = 50 bytes = 0.06 KB = 0.00 MB
  5. In the load summary, check the Rows duplicate/omitted number.

    This indicates the number of rows (if any) that tsload did not load because they did not satisfy the table constraints. A common cause of this would be a duplicate primary key. If tsload omitted any rows, review your CSV file, make the required adjustments, and then load it again.

    In the preceding example, tsload omitted two rows. This data needs to be cleaned and reloaded.

  6. After correctly loading your file, repeat this process to load data from any additional CSV files.

Loading data from an AWS S3 bucket

If you have data in .csv format stored in an AWS bucket, you can load it directly to ThoughtSpot.

Assigning S3 read-only role to your EC2 instance

If your cluster is on version 5.3.1 or later, you can assign an S3 read-only role to your ThoughtSpot EC2 instance(s) so the instance(s) can access the S3 bucket from which you want to load the data. This eliminates the need to enter the AWS S3 credentials when loading your data. For details, see: Using an IAM Role to Grant Permissions to Applications Running on Amazon EC2 Instances in Amazon’s AWS documentation.

If you are using S3 for persistent storage, and assigned the ec2rolewithfulls3access IAM role to your instance, you do not need to complete this step in order to load data with S3.

If you already configured the IAM role, you do not need to supply the --s3a_access_key, --s3a_secret_key and --s3a_region parameters when loading data with tsload.

To load data from an AWS S3 bucket, do the following:

  1. Log in to the Linux shell of your AWS instance using SSH.

  2. Use the following syntax to invoke tsload, specifying the appropriate flags and your data source file. If you do not supply the --s3a_bucket_name, tsload prompts you to enter it.

    $ tsload --source_file "/s3a/default/<my_file_in_aws>"
           --target_database "<my_database_in_ThoughtSpot>" --target_table "<my_table_in_the_database_in_ThoughtSpot>" --s3a_bucket_name <bucket_name>

    --source_file must contain the file path inside the bucket, and it must have the prefix /s3a/default.

    For example, to load from a file named directory/file.csv, the --source_file is /s3a/default/directory/file.csv.

    This example imports the CSV file teams.csv into the table teams in the database temp:

    $ tsload --source_file "/aws/default/teams.csv"
           --target_database "temp" --target_table "teams"
  3. After running the tsload command, you are prompted to enter additional AWS S3 information:

    • AWS S3a bucket name

      If you have no IAM role configured, you must enter the following information as well:

    • AWS S3a region

    • AWS S3a access key

    • AWS S3a secret key

    Optionally, these four pieces of information can be inserted at the beginning of the command (in step 2), using the following flags:

    • --s3a_bucket_name "<Name of bucket that contains the source CSV file>"

    • --s3a_region "<Region where the bucket is located>"

    • --s3a_access_key "<AWS S3 access key>"

    • --s3a_secret_key "<AWS S3 secret key>"

  4. After the processing begins, progress messages appear, and then source and load summary messages after the load is complete.

Loading data from a GCP GCS bucket

If you have data in .csv format stored in a GCS bucket, you can load it directly to ThoughtSpot.

Assigning GCS read-only role to your GCP instance

You can assign a read-only role to your ThoughtSpot GCP instance(s) so the instance(s) can access the GCS bucket from which you want to load the data. This eliminates the need to enter the GCP GCS credentials when loading your data. For details, refer to the storage setting detailed in the GCP create an instance section.

If you are using GCS for persistent storage, you already assigned the Set access for each API scope to your instance and specified Full storage access, so you do not need to complete this step in order to load data with GCS.

Create the database and table

  1. Log in to the Linux shell of your GCP instance using SSH.

  2. Invoke TQL:

     $ tql
    
     TQL>
  3. Create the database:

     TQL> CREATE DATABASE temp;
  4. Connect to the database:

     TQL> USE temp;
  5. Create the table

     TQL> create table teams (id int, name VARCHAR(255));
     TQL> exit;

Load data

Use the following syntax to invoke tsload, specifying the appropriate flags and your data source file:

$ tsload --source_file /gs/default/team.csv
       --target_database temp
       --target_table teams
       --bucket_name "my_gcs_bucket"
       --has_header_row 2>/dev/null

$ Header row read successfully
Source has 2 data rows, has header row, ignored row count 0
Waiting for rows to commit...(please wait)
Source summary
--------------
Data source:                 /gs/default/team.csv
Source data format           csv
Header row?                  yes
Tokenizer Options:           escape_char: "" field_separator: "," enclosing_char: "\"" null_value: "(null)" trailing_field_separator: false
Date format:                 %Y%m%d
Date time format:            %Y%m%d %H:%M:%S
Flexible mode?               no
Load summary
------------
Target table:                teams
Should empty target?         no
Status:                      Successful
Rows total:                  2
Rows successfully loaded:    2
Rows failed to load:         0
% of Rows successfully loaded: 100.00 %
Load Rate (MB/s):             0.00 MB/s
Load Rate (Rows/s):           1.13 Rows/s
Start time (Wallclock):       Wed Oct 30 23:30:11
End time (Wallclock):         Wed Oct 30 23:30:13
Total load time = 1.78 seconds = 0.03 minutes = 0.00 hours
Data size = 19 bytes = 0.02 KB = 0.00 MB

Verify the table contents

  1. Invoke TQL:

     $ tql
    
     TQL>
  2. Connect to the database:

     TQL> USE temp;
  3. Show the table data:

    TQL> select * from teams;

    Table contents are displayed:

    id|name
    -------
    1|sameer
    2|sandeep