Hue User impersonation for Presto doesn't work

[DRavikanth]

Is the issue already present in https://github.com/cloudera/hue/issues or discussed in the forum https://discourse.gethue.com? No

Describe the bug: After enabling Hue to impersonate user in Presto, observing

(pyhive.exc.OperationalError) Unexpected status code 401 Access Denied: Invalid credentials [SQL: select count(*) from event_detail] (Background on this error at: http://sqlalche.me/e/e3q8)

Steps to reproduce it?
options configured for SQLAlchemy in hue.ini

options='{"url": "presto://user:password@localhost:8443/hive/dwh","has_impersonation": true}'

Enabled impersonation in Presto

Hue version or source? (e.g. open source 4.5, CDH 5.16, CDP 1.0...). System info (e.g. OS, Browser...). Hue-4.8.0, open source Docker image. Safari on Mac OS Mojave

[romainr]

invalid credentials, please specify them https://docs.gethue.com/administrator/configuration/connectors/#presto

[DRavikanth]

Hi @romainr, It looks like my description is malformed. Please check the updated url now. I think it looks like a bug here. When impersonation is enabled, Logged in user is being replaced in the connection url, but the password is the one existing in the connection instead of the logged in user. I think its worth reopening this and taking a look. Thanks.

[romainr]

Impersonation means using user:password as the credentials but then performing the operations as the other users, so this sounds correct (the password of the logged in user is unknown here, or should be set toe ${PASSWORD} in the connection string cf. the docs (and this was not tested in a while, but it used to prompt the user for it then)

[DRavikanth]

When we create a Notebook for Presto via JDBC, it will pop up for the authentication. But, with SQLAlchemy, the user will not be prompted for password. I think this is the problem.

[DRavikanth]

@romainr, May I know the version of PyHive we are using for Hue? I tried finding but couldn't locate the version.

[romainr]

https://github.com/gethue/PyHive bu will probably rebase on https://github.com/dropbox/PyHive soon

[byungnam]

I also faced this error while enabling Trino connector "has_impersonation" option with interface=SqlAlchemy.

Luckily, this can be resolved by PyHive 0.6.3 principal_username option.

But then the problem now is, we cannot send the principal_username dynamically with current implementation. If we set a fixed principal_username in hue.ini file, then the whole user will be impersonated to that user. We can add self.options['connect_args']['principal_username'] = self.user.username at sql_alchemy.py, but then it will always try to impersonate even if you don't want to.

@romainr I think 'has_impersonation' option needs to be revised to use the above principal_username option. At least, we need to fix it as it is not working when password authentication is enabled.

[sbbagal13]

@byungnam I am also facing same issue for connecting Hue to Trino , If possible could you please share code snippet from hue.ini where you setup connection details for Trino

[byungnam]

@sbbagal13
Here is my hue.ini. Hope this can help you.

[notebook]
  [[interpreters]]
    [[[trino-hadoop-rabbit]]]
      name=Trino-Hadoop-Rabbit
      interface=sqlalchemy
      options='{"url":"trino://USER:[email protected]:8443/hadoop_rabbit/default", "connect_args":"{\"protocol\":\"https\"}", "has_impersonation": true}'

[sbbagal13]

@byungnam Thanks for responding, Somehow when I install only pyHive (0.6.3 or 0.6.5) I am getting trino dialect not found And if I add trino package on top of it , error coming as unrecognize parameter "has_impersonation" .
Also as mentioned in your earlier comment , how did you provided principal_username in it. I am using gethue/hue:4.10.0 docker image and Trino 393 .. Any suggestion to resolve the issue ?

[romainr]

👍 to make it smarter. AFAIK this is only used for Hive/Presto (and DaskSql):
https://github.com/cloudera/hue/blob/master/desktop/libs/notebook/src/notebook/connectors/sql_alchemy.py#L175

Also just rebased https://github.com/gethue/PyHive on https://github.com/dropbox/PyHive to get the Trino commit.