Known Issues in Hue

CDPD-40354: Uploading large files to WebHDFS from Hue secured using Knox fails

When you upload files larger than 300 MB to WebHDFS using Hue File Browser from a Hue instance that is secured using Knox, the upload fails and you see the following error on the Hue web interface: 502 bad gateway nginx error. This error occus only when you access the Hue instance from the Knox Gateway. Also, you may not find any exceptions in Hue or Knox service logs.

This issue is caused because the value of the addExpect100Continue is set to true. Therefore streaming large files causes out of memory error in Knox.

To resolve the issue, you must set the value of the addExpect100Continue to false.

SSH in to the Knox host as an Administrator.
Back up the following files:
- /opt/cloudera/parcels/CDH-[***VERSION***]/lib/knox/data/services/hue/1.0.0/service.xml
- /var/lib/knox/gateway/data/services/hue/1.0.0/service.xml
Open the /opt/cloudera/parcels/CDH-[***VERSION***]/lib/knox/data/services/hue/1.0.0/service.xml file for editing, search for the addExpect100Continue property, and set the value of this property to false as follows:
```
<param>
   <name>addExpect100Continue</name>
   <value>false</value>
</param>
```
Save the file and exit.
Open the /var/lib/knox/gateway/data/services/hue/1.0.0/service.xml file for editing, search for the addExpect100Continue property, and set the value of this property to false as follows:
```
<param>
   <name>addExpect100Continue</name>
   <value>false</value>
</param>
```
Save the file and exit.
Delete the deployment directory as follows:
```
rm -rf /var/lib/knox/gateway/data/deployments/
```
Restart the Knox service.

CDPD-58978: Batch query execution using Hue fails with Kerberos error

When you run Impala queries in a batch mode, you enounter failures with a Kerberos error even if the keytab is configured correctly. This is because submitting Impala, Sqoop, Pig, or pyspark queries in a batch mode launches a shell script Oozie job from Hue and this is not supported on a secure cluster.

There is no workaround. You can submit the queries individually.

CDPD-39330: Unable to use the pip command in CDP

You may not be able to use the pip command in CDP 7.1.7 and may see the following error when using pip in a command: “ImportError: cannot import name chardet”.

Follow the steps listed on Unable to use pip command in CDP.

CDPD-24294: Hue uses the unsafe-inline directive in its Content Security Policy (CSP) header

Hue 4 web interface uses the unsafe-inline directive in its CSP header. As a result, the application server does not set the CSP header in its HTTP responses, and therefore does not benefit from the additional protection against potential cross-site scripting issues and other modern application vulnerabilities which a properly configured CSP may provide. This could lead to application vulnerability.

This issue will be fixed in Hue 5. Until then, Cloudera recommends deploying additional security measures such as a firewall within the Hue server to control allowed connections, and SSO-based authentications mechanisms such as LDAP or SAML.

OPSAPS-61244: Cloudera Manager displays stale Hue configuration after upgrading to CDP 7.1.x from CDH 6.

After upgrading from CDH 6 to CDP 7.1.x, you may see stale configurations in Cloudera manager for the Hue service.

Manually restart the Hue service from Cloudera Manager.

ENGESC-9091: Setting idle session timeout for Hue does not work when the cluster is secured using Knox SSO

If Hue is configured with desktop.auth.backend.KnoxSpnegoDjangoBackend as the Authentication Backend, then the automatic idle session logout that is set by configuring the idle_session_timeout property does not take effect. You may also see 404 error while accessing Hue from the Knox UI when the idle_session_timeout property is not set to -1.

None

DOCS-10377: Hue UI is blank upon login after upgrading to CDP 7.1.7 from CDH 6

If your cluster was secured using Knox, and if you have upgraded from CDH 6 to CDP 7.1.7, then you may see a blank Hue screen. This could happen because the knox_proxyhosts parameter is newly introduced in CDP, and it is possible that this parameter is not configured in Cloudera Manager under Hue configuration.

Specify the host on which you have installed Knox in the Hue Knox Proxy Hosts configuration as follows:

Log in to Cloudera Manager as an Administrator.
Obtain the host name of the Knox Gateway by going to Clusters > Knox service > Instances.
Go to Clusters > Hue service > Configuration and search for the Knox Proxy Hosts field.
note
Cloudera Manager displays the following warning if the Knox Proxy Hosts field is empty when Knox Gateway is enabled on the CDP cluster: The parameter knox_proxyhosts cannot be empty. This can happen if there are no Knox Gateways. Please set the knox_proxyhosts to the list of hosts that have Knox Gateways.
Specify the Knox Gateway hostname in the Knox Proxy Hosts field.
Click Save Changes and restart the Hue service.

CDPD-18959: Downloading Impala query results containing special characters in CSV format fails with ASCII codec error

In CDP, Hue is compatible with Python 2.7.x, but the Tablib library for Hue has been upgraded from 0.10.x to 0.14.x, which is generally used with the Python 3 release. If you try to download Impala query results having special characters in the result set in a CSV format, then the download may fail with the ASCII unicode decode error.

To fix this issue, downgrade the Tablib library to 0.12.x.

SSH into the Hue server host.
Change directory to the following:
```
cd /opt/cloudera/parcels/CDH-7.x/lib/
```
Back up the hue directory:
```
cp -R hue hue_orginal
```
Change to the hue directory:
```
cd hue
```
Install the Wheel package using pip:
```
./build/env/bin/pip install wheel
```
The Wheel package is used to avoid recompiling your software during every install.
Install the Python Setuptools package for Hue as follows:
```
./build/env/bin/pip install setuptools==44.1.0
```

Install Tablib version 0.12.1 as follows:

./build/env/bin/pip install tablib==0.12.1

Go to Cloudera Manager and restart the Hue service.

OPSAPS-60196: Psycopg2 library needed for PostgreSQL-backed Hue when on RHEL 8 or Ubuntu 20 platforms

You may see a warning on the Host Inspector Results page stating that a compatible version of the Psycopg2 library is missing on your host if you have installed CDP 7.1.7 on RHEL 8 or Ubuntu 20 platforms and if you are using PostgreSQL as the backend database for Hue. This is because RHEL 8 and Ubuntu 20 contain Python 3 by default and Hue does not support Python 3.

Hue in Runtime 7 requires version 2.7.5 of the psycopg2 Python package for connecting to a PostgreSQL database. If you are on RHEL 8 or Ubuntu 20, then you must install one of the following compatible psycopg2 package versions before deploying CDP 7.1.7 on your cluster: 2.7.5, 2.7.6.1, and 2.7.7. For more information, see Installing the psycopg2 Python package for PostgreSQL-backed Hue.

OPSAPS-58927: Connection failed error when accessing the Search app (Solr) from Hue

If you are using Solr with Hue to generate interactive dashboards and for indexing data, and if you have deployed two Solr services on your cluster and selected the second one as a dependency for Hue, then Cloudera Manager assigns the hostname of the first Solr service and the port number of the second Solr service generating an incorrect Solr URL in the search section of the hue.ini file. As a result, you may see a “Connection failed” error when you try to access the Search app from the Hue web UI.

Log into Cloudera Manager as an Administrator.
Go to Clusters > Hue service > Configuration and add the following lines in the Hue Service Advanced Configuration Snippet (Safety Valve) for hue_safety_valve.ini field:
```
[search]
# URL of the Solr Server
solr_url=http://[***HOSTNAME***]:[***PORT***]/solr/
```
For example:
```
solr_url=http://solr2:4567/solr/
```
Click Save Changes.
Restart the Hue service.

CLR-72251: Invalid S3 URI error while accessing S3 bucket

The Hue Load Balancer merges the double slashes (//) in the S3 URI into a single slash (/) so that the URI prefix "/filebrowser/view=S3A://" is changed to "/filebrowser/view=S3A:/". This results in an error when you try to access the S3 buckets from the Hue File Browser through the port 8889.

The Hue web UI displays the following error: “Unknown error occurred”.

The Hue server logs record the “ValueError: Invalid S3 URI: S3A” error.

To resolve this issue, add the following property in the Hue Load Balancer Advanced Configuration Snippet:

Sign in to Cloudera Manager as an administrator.
Go to Clusters > Hue service > Configurations > Load Balancer and search for the Load Balancer Advanced Configuration Snippet (Safety Valve) for httpd.conf field.
Specify MergeSlashes OFF in the Load Balancer Advanced Configuration Snippet (Safety Valve) for httpd.conf field.
Click Save Changes.
Restart the Hue Load Balancer.
You should be able to load the S3 browser from both 8888 and 8889 ports.
Alternatively, you can use the Hue server port 8888 instead of the load balancer port 8889 to resolve this issue.

CLR-72255: Error while rerunning Oozie workflow

You may see an error such as the following while rerunning an an already executed and finished Oozie workflow through the Hue web interface: E0504: App directory [hdfs:/cdh/user/hue/oozie/workspaces/hue-oozie-1571929263.84] does not exist.

To resolve this issue, add the following property in the Hue Load Balancer Advanced Configuration Snippet:

Sign in to Cloudera Manager as an administrator.
Go to Clusters > Hue service > Configurations > Load Balancer and search for the Load Balancer Advanced Configuration Snippet (Safety Valve) for httpd.conf field.
Specify MergeSlashes OFF in the Load Balancer Advanced Configuration Snippet (Safety Valve) for httpd.conf field.
Click Save Changes.
Restart the Hue Load Balancer.

CDPD-16407: Python-psycopg2 package version 2.8.4 not compatible with Hue

Ubuntu 18.04 provides python-psycopg2 package version 2.8.4 but it is not compatible with Hue because of a bug in the Django framework.

Downgrade the package at the OS level by running the following command:

sudo apt install python-psycopg2==2.7.5

or install python-psycopg2 package using pip by running the following command:

sudo pip install psycopg2==2.7.5

DOCS-6344: Hue limitation after upgrading from CDH to CDP Private Cloud Base

The hive.server2.parallel.ops.in.session configuration property changes from TRUE to FALSE after upgrading from CDH to CDP Private Cloud Base. Current versions of Hue are compatible with this property change; however, if you still would like to use an earlier version of Hue that was not compatible with this property being FALSE and shared a single JDBC connection to issue queries concurrently, the connection will no longer work after upgrading.

Unsupported feature: Importing and exporting Oozie workflows across clusters and between different CDH versions is not supported

You can export Oozie workflows, schedules, and bundles from Hue and import them only within the same cluster if the cluster is unchanged. You can migrate bundle and coordinator jobs with their workflows only if their arguments have not changed between the old and the new cluster. For example, hostnames, NameNode, Resource Manager names, YARN queue names, and all the other parameters defined in the workflow.xml and job.properties files.

Using the import-export feature to migrate data between clusters is not recommended. To migrate data between different versions of CDH, for example, from CDH 5 to CDP 7, you must take the dump of the Hue database on the old cluster, restore it on the new cluster, and set up the database in the new environment. Also, the authentication method on the old and the new cluster should be the same because the Oozie workflows are tied to a user ID, and the exact user ID needs to be present in the new environment so that when a user logs into Hue, they can access their respective workflows.

CDPD-18491: PySpark and SparkSQL are not supported with Livy in Hue

Hue does not support configuring and using PySpark and SparkSQL with Livy in CDP Private Cloud Base.

INSIGHT-3707: Query history displays "Result Expired" message

You see the "Result Expired" message under the Query History column on the Queries tab for queries which were run back to back. This is a known behaviour.

None.