Compatibility among TigerGraph 3 versions

3.11.0

The graph field is now included in CDC messages generated by the TigerGraph CDC service.

Update any CDC message parsers or consumers to handle the new graph field.

3.11.0

In CDC messages, the format of map values has changed.

Modify applications processing CDC map values to match the new format.

3.10.0

New File Input Policy feature introduces a list of local files locations that may not be used to load data.

Review the file input/output policies before upgrading. Note that TigerGraph enforces an implicit blocklist of restricted file locations, which cannot be modified. See Input policy backward compatibility.

3.10.0

The gbar command is removed and replaced with gadmin backup.

Use gadmin backup for new backups. You can still restore older gbar backups with gadmin backup restore, since the restore process remains the same.

3.9.3

Edge operations require privileges on both the edge and its source/target vertices.

Grant the same data privilege on the edge type and the related vertex types.

3.9.3

PRINT now removes spaces from expression-based output labels. For example, count(DISTINCT @@ids) changes to count(DISTINCT@@ids).

Update queries if they depend on output labels with spaces.

3.9.2

Betweenness Centrality algorithm: reverse_edge_type changed to reverse_edge_type_set (SET<STRING>)

Update queries to use the new parameter name and type.

3.9.2

Vertices with string-type primary IDs cannot have empty string values.

Ensure input data has no empty string IDs.

3.9.2

Kafka Connector default mode changed from EOF="false" to EOF="true".

Set EOF explicitly in your configs if your workflow requires "false."

3.9.2

Default retention time for Informant and TS3 services reduced from 30 to 7 days.

Adjust retention settings for longer periods if needed.

3.9.2

/informant/metrics/get/cpu-memory accepts lists of ServiceDescriptors.

Update API calls to provide a list format.

3.9

UDFs named to_string() or with the tg_ prefix are not accepted due to new security checks.

Rename UDFs and update code to avoid these names for smooth upgrades.

3.6.3

A vertex having an empty string as its primary id is considered bad data and will not be loaded.

Check input files and loading jobs to insure you are not attempting to load vertices with no ID.

3.6

A single POST-ACCUM clause can no longer reference more than one vertex alias in Syntax V1.

Rewrite query, possibly using multiple `POST-ACCUM`s

3.6

GET /requesttoken endpoint is dropped.

Use POST /requestoken instead, with credentials in the request body.

3.6

Deploying TigerGraph in Kubernetes now requires more service account permissions than previous versions.

See Prerequisites section in kubernetes:quickstart-with-gke.adoc, kubernetes:quickstart-with-aks.adoc, and kubernetes:quickstart-with-eks.adoc.

3.4

The built-in role queryreader can no longer run queries that include updates to the database.

To emulate the old queryreader role, create a role with all queryreader privileges, and also grant the WRITE_DATA privilege to the new role.

3.4

The built-in role admin can no longer create users.

To emulate the old admin role, create a global role with all admin privileges, and also grant the WRITE_USER privilege to the new role.

3.2

Subqueries: A distributed mode main query cannot call a distributed mode subquery.

Refactor such queries.

3.2

Log files: Default logging level for GSQL logs has changed from DEBUG to INFO.

Set the logging level to GSQL to DEBUG when needed.

3.1.1

Update loading jobs which use TAGS

3.1.1

REST endpoint renamed from dbsanitycheck to deleted_vertex_check

Inform admin users

3.1

STRING COMPRESS data type is deprecated. New data objects may not have this data type. Existing data objects continue to work.

Use STRING data type instead for new schemas.

3.1

GSQL: A schema change can no longer drop a vertex type if it is used in an edge type.

Consider whether you intend to also remove the vertex type from any edges.