shield_lock
JinDaGe
← Back to skills
extension
Category: Development & EngineeringAPI key requirement unconfirmed

exasol-bucketfs

Exasol BucketFS file system management via exapump CLI. Covers listing, uploading, downloading, and deleting files and directories in BucketFS, BucketFS configuration, bucket structure, and use with UDFs.

personAuthor: jakexiaohubgithub
open_in_newRepositorydownloadDownload package

Exasol BucketFS Skill

BucketFS is Exasol's synchronous distributed file system: whatever is written to a bucket is replicated to every cluster node and mounted read-only inside UDFs at /buckets/<service>/<bucket>/<path>. Replication takes time, and recognised archives are extracted on top of it, so a fresh upload becomes usable only once synchronisation finishes. This skill covers moving files in and out of BucketFS and referencing them from scripts.

Routing Algorithm

Choose the narrowest matching route. If several apply, load all matching references before answering.

  1. Run a BucketFS operation — list, upload, download, delete, or configure the connection

    • Trigger phrases: exapump bucketfs, ls, cp, rm, upload, download, delete, --bfs-host, --bfs-bucket, bfs_write_password, config.toml, exapump profile add
    • Load: references/exapump-bucketfs-cli.md

  2. Stage a file for a UDF or SLC — JARs, pickled models, containers, and the SQL that points at them

    • Trigger phrases: upload jar, %jar, upload model, load model in UDF, upload SLC, activate container
    • Load: references/bucketfs-udf-usage.md

  3. Understand BucketFS itself — services, buckets, path forms, replication, durability

    • Trigger phrases: what is BucketFS, bfsdefault, service, bucket structure, replication, backup, atomic write, size limit
    • Load: references/bucketfs-concepts.md

An upload for a UDF usually needs routes 1 and 2 together: the command form comes from the CLI reference, the path spelling in the SQL from the usage reference.

Step 0: Establish Connection

Ensure a working exapump profile — BucketFS uses the profile's bfs_* fields — before running any command:

  1. If the user names a profile, test it with exapump bucketfs ls --profile <name>; always place --profile after the subcommand. Otherwise test the default profile with exapump bucketfs ls.
  2. On success, proceed — and keep the same --profile <name> after the subcommand on every later command, such as exapump bucketfs cp --profile <name> <local-file> <bucket-path>.
  3. On failure, run exapump profile list to see which profiles exist.
  4. If profiles exist, present them and ask which to use, then inspect that one with exapump profile show <name> — it masks credentials — to confirm the required fields are set. Never read, cat, or print ~/.exapump/config.toml; it stores credentials in clear text. If a credential value does appear in any command output, do not repeat it in the conversation.
  5. If no usable profile exists, have the user create one locally with exapump profile add <name> (omit --password and exapump prompts on a hidden line) or exapump profile init, then retry step 1. Never ask the user to paste a password into the conversation, and never pass one as a command-line argument.

Safety Rules

  • Before rm, show the exact bucket, path, and profile, then obtain explicit confirmation. BucketFS is not part of database backups, so a deletion is not recoverable from a database restore.
  • Before a cp that would overwrite an existing BucketFS or local file, inspect the target and obtain confirmation.
  • Never print profile passwords or pass them as command-line arguments when a profile can hold them securely. Inspect a profile with exapump profile show <name>, never by reading ~/.exapump/config.toml.

Related Skills

  • exasol-udfs: creating UDF scripts that read BucketFS files, and building the Script Language Containers staged here.
  • exasol-database: SQL-level operations and database connectivity.