TigerGraph DB Release Notes

TigerGraph Server 4.1.2 was released on December 20, 2024.

TigerGraph Server 4.1.1 was released on November 12, 2024.

TigerGraph Server 4.1.0 preview version was released on August 30, 2024.

TigerGraph 4’s feature set is an enhancement of what was available in 3.10.1. Therefore, features that were in 3.10.1 are not considered New Features.

Detailed List of New and Modified Features

Backup and Restore Enhancements

Security Enhancements

  • Fine-grained Privileges on Queries: gives adminstrators finer control over who may do what actions with which objects:

    • Allows privileges to be granted/revoked at the query level.

    • Adds new query privileges OWNERSHIP, INSTALL and EXECUTE, and splits WRITE to CREATE, UPDATE and DROP.

  • JWT Authentication: JWT authorization/authentication is more secure than the plaintext tokens used in earlier TigerGraph versions. Users can generate and use JWT tokens when sending REST requests to the RESTPP server [in 4.1.0] or GSQL server [in 4.1.2].

  • Password Complexity and Rotation Rules: Administrators can specify rules for the complexity of passwords and how long or how many times a password may be used, in order to improve security and help organizations to satisfy compliance requirements.

  • Expanded Audit Log Coverage: Audit logs now include gadmin activity, for more complete audit coverage.

  • [4.1.1] More detailed information in audit logs. Also, an authorized admin user can configure the audit logs to include PII.

  • New Built-in Role: globalobserver

    • The globalobserver role introduces read-only access for schemas and loading jobs on the global scope

High Availability (HA) Enhancements

  • Data Loading HA (Beta): If the database is replicated, this data loading mode offers faster loading, reduced disk usage, and HA with automatic failover. Available for TigerGraph’s data streaming connectors for cloud storage, Kafka, data warehouses, and Spark.

  • Change Data Capture (CDC) HA: CDC continues to function as long as at least one replica per partition is online. Previously, Replica 1 needed to remain operational.

  • [4.1.1] CDC messages have a "graph" field to specify which graph was affected.

  • [4.1.1] Cross-Region Replication now includes loading jobs.

  • Backup HA: Backups remain available even if a node fails. Previously, backups were unavailable if a node failed.

Data Streaming/Connector Enhancements

  • Malformed Loading Data Inspector: SHOW LOADING ERROR presents a sample of malformed data lines and their error types to an authorized user, enabling them to quickly locate and diagnose loading problems.

  • Data Export to AWS S3: Export query results in CSV format directly to AWS S3 buckets for efficient data sharing and analysis

  • Stream Data into Spark: Using the Spark data source reader API, users can run TigerGraph queries and stream the output to Spark as DataFrames. Streaming the output lifts the 2GB query response size limitation for non-streaming output.

  • Notification for Stuck Loader: If loading is stuck or acting abnormally, the loader sends an alert message to the console, including a diagnosis, enabling users to take timely action.

  • See also High Availability (HA) Enhancements

System Management and Monitoring Enhancements

Kubernetes Operator Enhancements

  • TigerGraph Operator 1.2.0 introduces several significant new features, including the ability to create services for sidecar containers, support for cluster storage resizing, new lifecycle hooks for TigerGraph CR, and enhanced Multi-AZ cluster resiliency.

    TigerGraph K8s Operator 1.2.0 details

  • [4.1.1] Added the namespace as a suffix to the HostName in the HostList configuration for TigerGraph on K8s.

  • [4.1.1] Support for customizing the external service port for the TigerGraph and sidecar listeners on K8s.

  • [4.1.1] Pre-upgrade hook in the TigerGraph Operator Helm chart to refresh the CRD automatically during the Operator upgrade.

Language Enhancements

  • REST APIs for Programmatic Use: Introduces a more complete and standard set of REST APIs:

    • provides coverage for database commands previously only available as GSQL commands.

    • allows developers and AI to more easily control and use TigerGraph programmatically.

  • [4.1.1] Added token function gsql_current_datetime() for loading jobs.

TigerGraph Suite

GraphStudio

  • Highlight Connected Edges When Hovering Over a Vertex:
    to show users the relationships and structure at a glance. Also available in Insights.

  • Collapsible Sidebar:
    to let users use their screen space more effiently.

  • Mandatory Password Change upon Expiration:
    A password must be changed when it expires or reaches a time/usage limit, enhancing security protocols. Not available on TigerGraph Cloud.

  • Customizable Naming of Reverse Edges:
    enables data modelers to apply more intuitive and domain-specific names.

TigerGraph Insights

  • Downloadable Query Output:
    as CSV or JSON

  • "Tree" View Respects Direction of Directed Edges:
    to depict hierarchical structures and dependencies more meaningfully

  • Support for Variables in Markdown Widget:
    for more context-aware and interactive dashboard displays

AdminPortal

  • Health Check Tool: The Health Check Tool in TigerGraph AdminPortal provides administrators with a comprehensive set of checks and diagnostics to ensure the system is running optimally.

  • Fine-grained Query Privileges in RBAC: AdminPortal UI for the fine-grained query privileges described above.

  • [4.1.2] Read-only Access to Logs: Permit users having the new prilivege APP_ACCESS_LOG (included in the globaldesigner and superuser built-in roles ) to read log files.

Fixed issues

Fixed and Improved in 4.1.2

Functionality

  • Fixed issue where query installation failed for single-node queries that initialize vertex set variables in conditional branches (e.g., if-else or case-when statements) (GLE-8908).

  • Fixed disk critical issue caused by the rebuilder getting stuck in a partitioned cluster after dropping vertex or edge attributes (CORE-4418).

  • Fixed issue where CDC errors were not properly recorded in logs unless the log level was set to DEBUG (CORE-4425).

  • Fixed query compilation error when iterating through a foreach statement to retrieve an accum type attribute in a query compiled with the -single or distributed keyword. Ensured operations between two different GroupByAccums hold the same data types and field names (GLE-9054).

  • Fixed restore failure when importing GSQL data due to excessive time taken by the GSQL warmup process (TP-6948).

  • Fixed an exception (NoSuchElementException) when removing all access tokens from the system (GLE-9022).

  • Fixed a CDC UNKNOWN ID issue occurring under high load in the GSE service (CORE-4444).

Improvements

  • Improved the performance of GSQL queries containing delete statements intended for deleting all vertices of a given type (GLE-8913).

  • Added new fields PodInitLabels and PodInitAnnotations in the PodTemplateSpec of the StatefulSet managing TigerGraph pods. These fields allow users to customize pod labels and annotations for initialization purposes (TP-6901).

  • Added functionality to allow Application Developers read-only access to TigerGraph logs (e.g., GSQL query logs, system logs) through GraphStudio without requiring sysadmin intervention (APPS-3253).

  • The GSQL REST endpoints now accept JWT tokens (GLE-8672).

Fixed and Improved in 4.1.1

Functionality

  • Fixed issue where local accumulators defined across multiple lines in a query were misinterpreted in the GSQL client (GLE-7833).

  • Fixed compile issue for escaped double quote inside string literal of loading job (GLE-8742).

  • Fixed issue where the input widget occasionally resets and loses content (APPS-3095).

  • Fixed inability to run gcollect on a k8s cluster (TP-6291).

  • Fixed Admin Portal log search when there are links to nonexistent logs (APPS-2874).

Crashes and Deadlocks

  • Fixed a stall in differential backups if the preceding full backup was created before any data was loaded into the system (CORE-3833).

  • Fixed issue where a loading job was stuck and couldn’t be cleared if it failed to start (TP-6419).

  • Fixed issue where the file loader enters an infinite loop if the source file contains only a header line (TP-6636).

Security

  • Fixed issue where users without any roles and privileges could access other users' information through the "/gsql/v1/users" endpoint (GLE-8477).

  • Fixed issue in file input/output policy where symbolic links in the allow/block list were unexpectedly dereferenced (GLE-7139).

  • Fixed issue where TG Cloud password could be changed without first validating the current password (APP-2829).

Known Issues and Limitations

Description Found In Workaround Fixed In

After a cluster expansion, the GPE service may remain in the warmup state.

4.1.1

If this happens, run gadmin restart gse -y to release the GPE.

TBD

When using Import All if the schema size in the .zip file is exceedingly large, the import may fail with an error messages like this:

Large catalog file key: /1/ReplicaList.json

3.2

TBD

If importing a role, policy, or function that has a different signature or content from the existing one, the one being imported will be skipped and not aborted.

For example:

  • If the original function is: create function lib1.func2(int param1, float param2, string param3) returns (bool) {}.

  • And the user imports the new function: create function lib1.func2(int param1) returns (bool) {}. This second one will be skipped.

3.10.0

Users need to re-create (delete and create) the imported role, policy, or function manually, and make sure that the importing one meets the requirements set by the existing one.

TBD

Row Policy (Preview Feature) does not yet filter or check vertex attribute data in upsert operations.

Such as,

  • A query with insert statements.

  • A file or Kafka loading job.

  • A DDL loading request.

  • Or a standard upsert request.

3.10.0

Users should restrict the access of creating/running queries and loading jobs for roles related to row policy.

TBD

In file INPUT and OUTPUT policy, if there exists 2 path (path1 and path2) in the configured policy list and path1 is parent path of path2, then path1 may not be effective.

3.2 and 3.10.0

Users should avoid using paths if they are nested.

For example, avoid this scenario, path2 = "/tmp/more" and path1= "/tmp".

3.10.1

When RESTPP sends a request to all GPEs, and if one is down, the request sent to it will timeout. Including the consistency_check request will also mark as timeout.

3.10.0

  1. Run /rebuildnow to rebuild all the segments.

    Running /rebuildnow when one gpe is down will result in the request timeout. This does not mean the request failed, instead only the currently running GPE will do the rebuild, and any rebuild requests sent to the down GPEs will result in a timeout.

  2. Run /data_consistency_check?realtime=false to check the consistency.

TBD

While running export graph if the disk space is not enough, or the data has not been detected, the export data will get stuck loading.

3.10.0

Restart all services in Admin Portal or the backend.

TBD

[tg_]ExprFunction.hpp will be automatically merged while importing single graphs. In some cases, query compilation may fail.

3.10.0

See Known Issues and Workarounds

TBD

Upgrading from a previous version of TigerGraph has known issues.

3.10.0

See section Known Issues and Workarounds for more details.

TBD

Input Policy feature has known limitations.

3.10.0

See section Input Policy Limitations for more details.

TBD

Change Data Capture (CDC) feature has known limitations.

3.10.0

See section CDC Limitations for more details.

TBD

If the FROM clause pattern is a multi-hop and the ACCUM clause reads both primitive and container type attributes or accumulators of a vertex, the internal query rewriting logic may generate an invalid rewritten output.

3.9.3

This results in the error message: It is not allowed to mix primitive types and accumulator types in GroupByAccum.

TBD

Users may see a high CPU usage caused by Kafka prefetching when there is no query or posting request.

3.9.3

TBD

TBD

GSQL query compiler may report a false error for a valid query using a vertex set variable (e.g. Ent in reverse_traversal_syntax_err) to specify the midpoint or target vertex of a path in a FROM clause pattern.

TBD

TBD

TBD

If a loading job is expected to load from a large batch of files or Kafka queues (e.g. more than 500), the job’s status may not be updated for an extended period of time.

3.9.3

In this case, users should check the loader log file as an additional reference for loading status.

TBD

When a GPE/GSE is turned off right after initiating a loading job, the loading job is terminated internally. However, users may still observe the loading job as running on their end.

3.9.3

Please see Troubleshooting Loading Job Delays for additional details.

TBD

For v3.9.1 and v3.9.2 when inserting a new edge in GPR and INTERPRET mode, the GPE will print out a warning message because a discriminator string is not set for new-inserted edges. Creating an inconsistent problem in delta message for GPR and INTERPRET mode.

3.9.2

Please see Troubleshooting Loading Job Delays for additional details.

3.9.3

GSQL EXPORT GRAPH may fail and cause a GPE to crash when UDT type has a fixed STRING size.

TBD

TBD

TBD

After a global loading job is running for a while a fail can be encountered when getting the loading status due to KAFKASTRM-LL not being online, when actually the status is online. Then the global loading process will exit and fail the local job after timeout while waiting the global loading job to finish.

TBD

TBD

TBD

When the memory usage approaches 100%, the system may stall because the process to elect a new GSE leader did not complete correctly.

TBD

This lockup can be cleared by restarting the GSE.

TBD

If the CPU and memory utilization remain high for an extended period during a schema change on a cluster, a GSE follower could crash, if it is requested to insert data belonging to the new schema before it has finished handling the schema update.

TBD

TBD

TBD

When available memory becomes very low in a cluster and there are a large number of vertex deletions to process, some remote servers might have difficulty receiving the metadata needed to be aware of all the deletions across the full cluster. The mismatched metadata will cause the GPE to go down.

TBD

TBD

TBD

Subqueries with SET<VERTEX> parameters cannot be run in Distributed or Interpreted mode.

TBD

(Limited Distributed model support is added in 3.9.2.)

TBD

Upgrading a cluster with 10 or more nodes to v3.9.0 requires a patch.

3.9

Please contact TigerGraph Support if you have a cluster this large. Clusters with nine or fewer nodes do not require the patch.

3.9.1

Downsizing a cluster to have fewer nodes requires a patch.

3.9.0

Please contact TigerGraph Support.

TBD

During peak system load, loading jobs may sometimes display an inaccurate loading status.

3.9.0

This issue can be remediated by continuing to run SHOW LOADING STATUS periodically to display the up-to-date status.

TBD

When managing many loading jobs, pausing a data loading job may result in longer-than-usual response time.

TBD

TBD

TBD

Schema change jobs may fail if the server is experiencing a heavy workload.

TBD

To remedy this, avoid applying schema changes during peak load times.

TBD

User-defined Types (UDT) do not work if exceeding string size limit.

TBD

Avoid using UDT for variable length strings that cannot be limited by size.

TBD

Unable to handle the tab character \t properly in AVRO or Parquet file loading. It will be loaded as \\t.

TBD

TBD

TBD

If System.Backup.Local.Enable is set to true, this also enables a daily full backup at 12:00am UTC.

3.9.0

TBD

3.9.1

The data streaming connector does not handle NULL values; the connector may operate properly if a NULL value is submitted.

TBD

Users should replace NULL with an alternate value, such as empty string "" for STRING data, 0 for INT data, etc. (NULL is not a valid value for the TigerGraph graph data store.)

TBD

Automatic message removal is an Alpha feature of the Kafka connector. It has several known issues.

TBD

TBD

TBD

The DATETIME data type is not supported by the PRINT … TO CSV statement.

3.9.0

TBD

3.9.1

The LDAP keyword memberOf for declaring group hierarchy is case-sensitive.

3.9

Check the case of the keywords for memberOf. This has been fixed in versions 3.10.1 and above.

3.10.1

Compatibility Issues

Description Version Introduced

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

v4.1.1

A full export package now includes access policies and template queries.

v4.1.0

Users could encounter file input/output policy violations when upgrading a TigerGraph version. See Input policy backward compatibility.

v3.10.0

When a PRINT argument is an expression, the output uses the expression as the key (label) for that output value. To better support Antlr processing, PRINT now removes any spaces from that key. For example, count(DISTINCT @@ids) becomes count(DISTINCT@@ids).

v3.9.3+

Betweenness Centrality algorithm: reverse_edge_type (STRING) parameter changed to reverse_edge_type_set (SET<STRING>), to be consistent with edge_type_set and similar algorithms.

v3.9.2+

For vertices with string-type primary IDs, vertices whose ID is an empty string will now be rejected.

v3.9.2+

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

v3.9.2+

The default retention time for two monitoring services Informant.RetentionPeriodDays and TS3.RetentionPeriodDays has reduced from 30 to 7 days.

v3.9.2+

The filter for /informant/metrics/get/cpu-memory now accepts a list of ServiceDescriptors instead of a single ServiceDescriptor.

v3.9.2+

Some user-defined functions (UDFs) may no longer be accepted due to increased security screening.

  • UDFs may no longer be called to_string(). This is now a built-in GSQL function.

  • UDF names may no longer use the tg_ prefix. Any user-defined function that began with tg_ must be renamed or removed in ExprFunctions.hpp.

v3.9+

Deprecations

Description Deprecated Removed

The format for tuple structures in CDC messages will change in a future veresion. The future format is likely to be similar to the new format for maps.

4.1.1

TBD (possibly 4.2)

The use of plaintext tokens in authentication is deprecated. Use OIDC JWT Authentication instead.

3.10.0

TBD

Vertex-level Access Control (VLAC) and VLAC Methods are removed and are no longer available.

3.10.0

4.1.0

Spark Connection via JDBC Driver is now deprecated and will no longer be supported.

3.10.0

TBD

Build Graph Patterns is deprecated and will not be updated or supported and instead we are focusing on Insights as the tool of choice for building visual queries.

v3.9.3

TBD

Kubernetes classic mode (non-operator) is deprecated.

v3.9

TBD

The WRITE_DATA RBAC privilege is deprecated.

v3.7

TBD

Release notes for previous versions