Analytics/Systems/Kerberos/UserGuide

From Wikitech
Jump to navigation Jump to search

Purpose of Kerberos

Hadoop by default does not ship with a strong authentication mechanism for users and daemons. In order to enable its "Secure mode", an external authentication service must be plugged in, and the only compatible one is Kerberos.

When enabled, it means that users and daemons will need to authenticate to our Kerberos service before being able to use Hadoop. Please read the next sections to get more info about what do to.

High level overview


Analytics Kerberos infra scheme.svg


The picture depicts a high level overview of how Kerberos authentication affects users. First of all, it must be noticed that the Hadoop cluster is the only part of the infrastructure that will be configured to use Kerberos. The red lines show what are the parties that will require to authenticate to Kerberos to be able to use Hadoop:

  • Druid, since its deep storage is Hadoop HDFS. Please note that this will not mean that Druid itself will require Kerberos authentication from users, but only that Druid itself will need to authenticate before fetching data from HDFS. This means that Superset and Turnilo dashboards will keep working as before, without changes.
  • Users logging on stat100[4,5,7] and notebook100[3,4]. If anybody wants to use any tool that interacts with Hadoop (Oozie, Hive, Spark, Notebooks, etc..) they will need to authenticate via Kerberos.

Will my files on HDFS change when Kerberos is enabled?

Nothing will change, including file permissions. Kerberos is something that will allow the user to give a proof of identify for its username before accessing/modifying a file, nothing more.

How do I..

Authenticate via Kerberos

Run the kinit command, insert your password and then execute any command (spark, etc..). This is very important since if you don't do it, you'll see horrible error messages reported by basically anything you'll use. The kinit command grants you a so called Kerberos TGT (Ticket Granting Ticket), that will be used to allow you to authenticate to various services/hosts. The ticket last 24h, so you will not need to run kinit every time, just once a day. You can inspect the status of your ticket via klist.

Get a password for Kerberos

Please request an identity in https://phabricator.wikimedia.org/T237605. If you have any doubt, feel free to ask to the Analytics team on Freenode #wikimedia-analytics or via email. You'll receive an email containing a temporary password, that you'll be required to change during you first authentication (see section above).

FAQ:

  • This is really annoying, can't we just use LDAP or something similar to avoid another password?
    • We tried really hard but for a lot of technical reasons, the integration would be complicated and cumbersome to maintain for Analytics and SRE. There might be some changes in the future, but for now we'll have to deal with another password to remember.

Run a recurrent job via Cron or similar without kinit every day

The option that is currently available is a Kerberos keytab: a file with permissions set that only the owner can read, holding the password to authenticate to Kerberos. We use keytabs for daemons/services, and we'll plan to provide those to users with the need to run periodical jobs. The major drawbacks are:

  • our security standard lowers down a bit, since it is sufficient to ssh to a host to access HDFS (as opposed to also know a password). This is more or less the current scheme, so not a big deal, but we have to think about it.
  • The keytab needs to be generated for every host that needs to have this automation and it also needs to be regenerated and re-deployed when the user changes the password (this doesn't happen for daemons of course). It is currently not automated, and it requires a ping to Analytics every time..

The solution that we are currently working on is the following: every user of the analytics-privatedata-users POSIX group will be able to sudo/impersonate the analytics-privatedata user, that in turn will be able to read a keytab on some hosts (likely stat1007 and notebook1003) and authenticate to Kerberos without password. In order to hide complexity and help users we created the kerberos-run-command tool:

# No kinit done, hence no credentials for my user
elukey@an-tool1006:~$ klist
klist: No credentials cache found (filename: /tmp/krb5cc_13926)

# As expected, a simple ls will fail
elukey@an-tool1006:~$ hdfs dfs -ls /
[..cut..]
java.io.IOException: Failed on local exception: java.io.IOException: javax.security.sasl.SaslException: GSS initiate failed [Caused by GSSException: No valid credentials provided (Mechanism level: Failed to find any Kerberos tgt)]; Host Details : local host is: "an-tool1006.eqiad.wmnet/10.64.5.32"; destination host is: "analytics1029.eqiad.wmnet":8020;
[..cut..]

# First attempt to use the kerberos-run-command an-tool1006
elukey@an-tool1006:~$ kerberos-run-command analytics-privatedata hdfs dfs -ls /
The user keytab that you are trying to use (/etc/security/keytabs/analytics/analytics-privatedata.keytab) doesn't exist or it isn't readable from your user, aborting...

# The problem is that the keytab is not readable by all
# analytics-privatedata-users members directly, but they
# have to sudo first:
elukey@an-tool1006:~$ sudo -u analytics-privatedata kerberos-run-command analytics-privatedata hdfs dfs -ls /
Found 5 items
drwxr-xr-x   - hdfs hadoop          0 2019-06-20 06:00 /system
drwxrwxrwt   - hdfs hdfs            0 2019-11-14 15:03 /tmp
drwxr-xr-x   - hdfs hadoop          0 2019-10-25 16:24 /user
drwxr-xr-x   - hdfs hdfs            0 2019-01-17 13:42 /var
drwxr-xr-x   - hdfs hadoop          0 2019-06-25 13:46 /wmf

# It worked! Interesting question: does my user have credentials now? Let's check...
elukey@an-tool1006:~$ klist
klist: No credentials cache found (filename: /tmp/krb5cc_13926)

# Why not? Because only the analytics-privatedata user has:
elukey@an-tool1006:~$ sudo -u analytics-privatedata klist
Ticket cache: FILE:/tmp/krb5cc_498
Default principal: analytics-privatedata/an-tool1006.eqiad.wmnet@WIKIMEDIA

Valid starting       Expires              Service principal
11/14/2019 15:03:33  11/15/2019 01:03:33  krbtgt/WIKIMEDIA@WIKIMEDIA
	renew until 11/15/2019 15:03:33

# This may seem confusing at first, but it makes sense, since we had to sudo
# to be able to read the keytab.
# Corollary: the analytics-privatedata user is not a replacement for your
# kerberos authentication, only a convenient way to run recurrent jobs via
# cron or similar.

Check the Yarn Resource Manager's UI

Nothing changed for the Yarn's UI!

Check Hue

Nothing changed for Hue!

Use Hive

The hive cli is compatible with Kerberos, even if it uses an old protocol (connecting to the Hive Metastore and HDFS directly). The beeline command line uses the Hive 2 server via JDBC and it is also compatible with Kerberos. You just need to authenticate as described above and then run the tool on an-tool1006.eqiad.wmnet.

Use Spark 2

On stat100[4,5,7] and notebook100[3,4] authenticate via kinit and then use the spark shell as you are used to. There are currently some limitations:

  • spark2-thriftserver requires the hive keytab, that is only present on an-coord1001, so when running on client nodes it will return the following error: org.apache.hive.service.ServiceException: Unable to login to kerberos with given principal/keytab

Use Jupyterhub (SWAP replica)

You can authenticate to Kerberos running kinit in the Terminal window. Please remember that it will be needed only once every 24h, not every time.

Use Hive2 actions in Oozie

+    <credentials>
+        <credential name='my-hive-creds' type='hive2'>
+            <property>
+                <name>hive2.server.principal</name>
+                <value>hive/an-coord1001.eqiad.wmnet@WIKIMEDIA</value>
+            </property>
+            <property>
+                <name>hive2.jdbc.url</name>
+                <value>jdbc:hive2://an-coord1001.eqiad.wmnet:10000/default</value>
+            </property>
+        </credential>
+    </credentials>

     <start to="aggregate"/>

-    <action name="aggregate">
+    <action name="aggregate" cred="my-hive-creds">

Please note: the place in which you put credentials in a workflow is not arbitrary, but it must follow the worflow's oozie schema (like https://oozie.apache.org/docs/3.3.1/WorkflowFunctionalSpec.html#Oozie_Schema_Version_0.4).

Update Spark actions in Oozie

+    <credentials>
+        <credential name="hcat-cred" type="hcat">
+            <property>
+                <name>hcat.metastore.principal</name>
+                <value>hive/an-coord1001.eqiad.wmnet@WIKIMEDIA</value>
+            </property>
+            <property>
+               <name>hcat.metastore.uri</name>
+               <value>an-coord1001.eqiad.wmnet:9083</value>
+            </property>
+        </credential>
+    </credentials>


    <start to="generate_restbase_metrics"/>

-    <action name="generate_restbase_metrics">
+    <action name="generate_restbase_metrics" cred="hcat-cred">
        <spark xmlns="uri:oozie:spark-action:0.1">

Please note: the place in which you put credentials in a workflow is not arbitrary, but it must follow the worflow's oozie schema (like https://oozie.apache.org/docs/3.3.1/WorkflowFunctionalSpec.html#Oozie_Schema_Version_0.4).

How do I... (Analytics admins version)

Check the status of the HDFS Namenodes and Yarn Resource Managers

Most of the commands are the same, but of course to authenticate as the user hdfs you'll need to use a keytab:

sudo -u hdfs kerberos-run-command hdfs /usr/bin/yarn rmadmin -getServiceState an-master1001-eqiad-wmnet

sudo -u hdfs kerberos-run-command hdfs /usr/bin/hdfs haadmin -getServiceState an-master1002-eqiad-wmnet