Application Services | Backup and Recovery | Dashboard (UI | File System | Frames | Hadoop | High Availability (HA) | Hive | Jupyter | MLRun | Logging and Monitoring Services | Nuclio | Trino | Security | Spark | TSDB | User Management | Web APIs
Managed Application Services
| ID |
Description |
Workaround |
Opened in |
| IG-14150 |
A user whose username was previously in use in the platform cannot run application services.
|
NA |
v2.5.0 |
| IG-21024 |
GET mlrun/api/runs... fail with "RemoteDisconnected('Remote end closed connection without response') |
NA, dependant on NGINX upgrade. |
v3.5.0 |
Jupyter Notebook
| ID |
Description |
Workaround |
Opened in |
| IG-11174 |
Running Scala code from a Jupyter notebook isn't supported in the current release.
|
Run Scala code by using spark-submit, or use Python. |
v2.0.0 |
| NA |
Jupyter error "Server Connection Error: A connection to the Jupyter server could not be established...." Root cause: The Iguazio session expired. |
Supply the iguazio session credentials. |
NA |
V3IO Frames
| ID |
Description |
Workaround |
Opened in |
| 21993 |
When starting to ingest to a new stream in a “round-robin” mode, the first message starts from zero. |
NA |
v3.5.3 |
Frames NoSQL Backend ("nosql"/"kv")
| ID |
Description |
Workaround |
Opened in |
| IG-16167 |
The write client method's "overwriteTable" save mode isn't supported for partitioned tables. [Tech Preview]
|
NA |
v2.0.0 |
MLRun
| ID |
Description |
Workaround |
Opened in |
| ML-211 |
The default-project directory (default) is shared for read-only and execution.
|
NA |
v1.3.0 |
| IG-16531 |
Beginning with platform version 3.0.0 / MLRun version 0.6.0, project-name restrictions are enforced for MLRun projects both from the dashboard (UI) and the MLRun API. Project names must conform to the RFC 1123 DNS label-name requirements (DNS-1123 label): 1–63 characters, contain only lowercase alphanumeric characters (a–z, 0–9) or hyphens (-), and begin and end with a lowercase alphanumeric character. Older MLRun projects whose names don't conform to these restrictions will be in a degraded mode and won't have Nuclio functionality (no serverless functions or API gateways).
|
Replace the old projects (you cannot simply rename them). For assistance in upgrading old projects, contact Iguazio's support team. |
v3.0.0 |
Nuclio Serverless Functions
| ID |
Description |
Workaround |
Opened in |
| IG-13134 |
Function deployment for the Ruby, Node.js, and .NET Core runtimes isn't supported for platform installations without an internet connection (offline installations).
|
NA |
v2.0.0 |
| IG-12487 |
In case of a successful automatic deployment of a Nuclio function for a request that previously timed out — for example, if a requested resource, such as a GPU, becomes available after the time-out failure — the function status in the dashboard is still Error.
|
NA |
v2.3.0 |
Trino and Hive
| ID |
Description |
Workaround |
Opened in |
| IG-20818 |
Spill to disk is supported with Hive connectors only.
|
NA |
v3.4.2 |
| IG-11757 |
You can't change the MySQL DB user for an existing Hive MySQL DB.
|
Either reconfigure Hive for Trino to set a different DB path that doesn't yet exist (in addition to the different user); or disable Hive for Trino, apply your changes, delete the current MySQL DB directory (using a file-system command), and then re-enable Hive for Trino and configure the same DB path with a new user. |
v2.2.0 |
| IG-15849 |
Invalid Trino configuration files or connector definitions might cause the Trino service and all dependent services to fail. This is the expected behavior.
|
Delete the problematic configuration or manually revert relevant changes to the default Trino configuration files, save the changes to the Trino service, and select Apply Changes on the dashboard Services page. |
v2.8.0 |
| NA |
In rare cases, after failing to deploy incorrect edits to a default configuration file, it might not be possible to revert the default Trino configuration from the dashboard. |
NA |
NA |
| IG-18009 |
Running the Hive CLI from a Jupyter Notebook service when there's no web-shell service in the cluster might fail if another user had previously run the CLI from another instance of Jupyter.
|
Ensure that there is at least one web-shell service in the cluster. |
v3.2.0 |
Spark
Spark UI
| ID |
Description |
Workaround |
Opened in |
| IG-12143 |
The Spark UI might display broken links after killing a running application from the UI.
|
Close the current Spark UI browser tab and reopen the UI in a new tab. |
v2.3.0 |
Spark Streaming
| ID |
Description |
Workaround |
Opened in |
| IG-6471 |
Spark streaming app doesn't consume from 'new' shards when increasing shards amount.
|
Restart the Spark Streaming consumer application. |
v1.5.0 |
Spark Operator
| ID |
Description |
Workaround |
Opened in |
| IG-20546 |
Remote Spark fails to ingest from Azure with error similar to: 2022-04-10 12:52:54,217 [info] run executed, status=error runtime error: </br>An error occurred while calling o52.load. : java.lang.NoClassDefFoundError: org/eclipse/jetty/util/ajax/JSON$Convertor at org.apache.hadoop.fs.azure.NativeAzureFileSystem.createDefaultStore(NativeAzureFileSystem.java:1404) at org.apache.hadoop.fs.azure.NativeAzureFileSystem.initialize(NativeAzureFileSystem.java:1331) at org.apache.hadoop.fs.FileSystem.createFileSystem(FileSystem.java:3303)
|
Add these parameters to the Spark conf (adds the same set of jars to both the driver and the executor): conf = (SparkConf() .set("spark.driver.extraClassPath","/hadoop/share/hadoop/tools/lib/hadoop-azure-3.2.0.jar:/hadoop/share/hadoop/tools/lib</br> /azure-storage-7.0.0.jar:/hadoop/share/hadoop/hdfs/lib/jetty-util-ajax-9.3.24.v20180605.jar:/hadoop/share/hadoop/hdfs/</br> lib/jetty-util-9.3.24.v20180605.jar") .set("spark.executor.extraClassPath", "/hadoop/share/hadoop/tools/lib/hadoop-azure-3.2.0.jar:/hadoop/share/hadoop/tools/lib</br> /azure-storage-7.0.0.jar:/hadoop/share/hadoop/hdfs/lib/jetty-util-ajax-9.3.24.v20180605.jar:/hadoop/share/hadoop/hdfs/</br> lib/jetty-util-9.3.24.v20180605.jar")) |
v3.4.0 |
Web APIs
Simple-Object Web API
| ID |
Description |
Workaround |
Opened in |
| IG-3453 |
GET Object and HEAD Object don't return correct ETag and Content-Type metadata.
>}} |
NA |
v1.0.0 |
| IG-13449 |
A range request with an out-of-bounds range returns HTTP status code 200 instead of 400.
>}} |
NA |
v2.5.0 |
NoSQL Web API
| ID |
Description |
Workaround |
Opened in |
| IG-17963 |
Range-scan requests following recreation of a data container or a container directory might return a 513 error.
|
Reissue the request. |
v3.0.0 |
Streaming Web API
| ID |
Description |
Workaround |
Opened in |
| IG-4310 |
PutRecords with a stream path that points to an existing non-stream directory returns error ResourceNotFoundException instead of ResourceIsnotStream.
|
NA |
v1.0.0 |
| IG-4310 |
UpdateStream with an invalid request-body JSON format may not return the expected InvalidArgumentException error.
|
NA |
v1.0.0 |
Logging and Monitoring Services
| ID |
Description |
Workaround |
Opened in |
| IG-9029 |
Data-cluster statistics are not available when the platform is in offline mode.
|
NA |
v3.0.0 |
Hadoop
| ID |
Description |
Workaround |
Opened in |
| NA |
The following Hadoop (hdfs) commands aren't supported: createSnapshot, deleteSnapshot, getfattr, setfattr, setfacl, and setrep.
|
NA |
v1.0.0 |
File System
| ID |
Description |
Workaround |
Opened in |
| IG-3605 |
File-system operations are not supported for stream shards.
|
NA |
v1.5.0 |
Security and User Management
| ID |
Description |
Workaround |
Opened in |
| IG-17295 |
Changing the token-expiration period for the authenticator service (id_tokens_expiration configuration) doesn't affect existing tokens.
|
Contact Iguazio's support team. |
v3.0.0 |
| IG-21856 |
Attempting to synchronize users who are members of a large number of OUs without specifying a group filter may cause the IDP service to crash, causing the synchronization job to hang. |
Use a group filter. |
v3.4.2 |
Backup, Recovery, and High Availability (HA)
| ID |
Description |
Workaround |
Opened in |
| IG-13840 |
In Azure cloud environments, in case of failure in the main master application node of a Kubernetes application cluster, the platform's access to the application nodes is lost and it can no longer manage the cluster's application services. Previously run application services will continue to run but cannot be stopped, and new services cannot be created.
|
NA |
v2.1.0 |
| IG-6475 |
The GetItems NoSQL Web API operation might fail during failover.
|
NA |
v1.5.3 |
Dashboard (UI)
Note
See the
MLRun and
Nuclio known issues for dashboard issues related specifically to these frameworks.
| ID |
Description |
Workaround |
Opened in |
| IG-22403 |
Possible errors on SSO login: |
|
|
|
Account Already Exists: Your Security Admin may have changed the provider, and a stale instance of your user record still exists. |
Contact your Security Admin to remove your user from Keycloak (only) and log in again. |
v3.5.5 |
|
500 Internal server error: This occurs when the Oauth2 application refused to acknowledge your current cookie session. |
Clear your browser cookies (_oauth2_cookie) and try again. |
v3.5.5 |
Notes