zookeeper error during installation in Install Path A

[#1]

Hello All,

New to pretty much everything here, so please excuse my rookie-level explanation of my issue.

I've tried twice now to prepare a Cloudera instance using different sets of hardware for each attempt. Each time I attempt to complete the installation, I'm met with this zookeeper error (when it tries to run $> zookeeper/zk.sh ["init","--myid=1","ignoreExists"]:

Error found before invoking supervisord: 'getpwnam(): name not found: zookeeper'

As a result, nothing else starts. 

The screen I'm working from is the "Cluster Setup - First Run Command" using Cloudera Manager. I've configured the "cloudera-scm" user as I don't run anything as root using this guide:

https://www.cloudera.com/documentation/enterprise/latest/topics/install_singleuser_reqts.html#xd_583...

I'm following Install Path A using the guide here:

https://www.cloudera.com/documentation/enterprise/latest/topics/cm_ig_install_path_a.html

Near as I can tell, this is because the zookeeper user does not exist. I don't recall seeing a place to create this user during the install. I have saved all screenshots from my installation so I can reference those if you need them.

Any help is appreciated! Thanks!

[#2]

 me too. Help please.

[#3]

I meet the same problem. Do you solve it?

[#4]

Could you please let us know the

hardware , cdh version ,os details.

are you using cloudera manager to setup zookeeper.

[#5]

Greetings,

I'm having the very same issue. 

Tried creating the account locally, but then the error message changes to Exception: Non-root agent cannot execute process as user 'zookeeper'.

The deployment is a test environment, using a Linux Ubuntu 14.04 LTS VM, and CDH 5.11.

Any help will be greatly appreciated.

Tahnks,

Radu

[#6]

Update:

if the user/group (zookeeper/zookeeper) exist, the lines of code throwing the exception are:

user = self.raw["user"]
group = self.raw["group"]
uid, _ = user_group_to_ids(user, group)
if os.getuid() != 0 and os.getuid() != uid:
raise Exception("Non-root agent cannot execute process as user '%s'" % user)

The sudoers file is updated for both user and group:

# Members of the admin group may gain root privileges
%admin ALL=(ALL) ALL
%zookeeper ALL=(ALL) NOPASSWD: ALL

# Allow members of group sudo to execute any command
%sudo ALL=(ALL:ALL) ALL

# See sudoers(5) for more information on "#include" directives:

#includedir /etc/sudoers.d
radu ALL=(ALL) NOPASSWD: ALL
cloudera-scm ALL=(ALL) NOPASSWD: ALL
zookeeper ALL=(ALL) NOPASSWD: ALL

What could be the casue?

Thanks,

Radu

[#7]

Hi Radu, Have you resolved the problem? I'm encountering the similar problem when using cloudera manager 5.11. I'm transferring form non-HA to HA, then it threw an exception that "Non-root agent cannot execute process as user 'hdfs'", and then the error "Journal Storage Directory /data/d1/dfs/jn/scorpio not formatted". Thanks, MH

[#8]

Nope, haven't been able to solve it. Didn't spend too much time on it
though, as I'm samped with something else at the moment.

Thanks,
Radu

[#9]

Since I dont know the exact steps that you followed I am going to throw my guess here see if that will solve the issue . 

Did you check your firewall status .? 

What user are you runining while firing the cloudera manager install . ?

sometimes the intial run by cloudera manager fails due to poor hardware  / internet connectivity 

[#10]

Thanks for your reply. The firewall is disabled, and the manager is
installed using the root account. When deploying cdh, i'm using a local
accout, other than the zookeeper...

Thanks,
Radu

[#11]

if i understant correctly you are trying to add service - zookeeper in your cluster using cloudera manager web ui ? 

correct me if i am wrong  

does your local account has full administrative rights under cm ? 

[#12]

So, what I'm trying to do is to deploy a new cluster from the Manager. And when it gets to the zookeeper, I'm getting those errors.

Yes, the account I'm using is part of root and sudo groups and I've enabled sudo passwordless execution for it.

My question is, why is the setup looking for the ZooKeeper acocunt? I haven't found anything stated in the documentation about this.

thanks,

Radu

[#13]

Hi Radu,

Did you find the solution for it. I am also facing the same problem.

This problem has persisted for more than two months now, but still has no solution.

Regards,

Satish

[#14]

Hi Satish,

Nope, not been able to solve it yet. This is becoming frustrsting as the
install path is simply using their own installer, pushed from the manager...

Thanks,
Radu

[#15]

"My question is, why is the setup looking for the ZooKeeper acocunt? I haven't found anything stated in the documentation about this. " 

Obviously you would have choose that parcels to be deployed during your installation. 

Now give us some logs like Cloudera-scm-server , cloudera-scm-agent to help you . 

[#16]

Hello,

Thanks for your reply. There is a limitation for 50000 characters per insert code - is there a way to upload the log files as attachments?

Thanks,

Radu

[#17]

Would you be able to perform a tail -f command and post the logs

also this post is simliar to your issue i dont know if that is going to help you meantime .

they there is a internal jira . Not sure if they fixed it .

I usually dont choose single user mode when i go for Path A installation for POC or testing .

http://community.cloudera.com/t5/Cloudera-Manager-Installation/Cloudera-Single-User-Mode-installatio...

[#18]

Hi,

I have same problem. I'm using the cloudera manager to install CDH 5.7.2 and I'm finishing the installation

but I have this error during the "first run command" step

I added the zookeeper user and group to test into sudoers file.. but I did't have success with this error.

I'm using single user mode during the installation and RedHat 7

I did the ps and I have this services running with supervisor

I have no idea.. how to solve.

Thank you a lot

[#19]

Your screenshots are not visible.

[#20]

 Please, are there any solution for this issue? I have the same problem with CDH 5.11. thanks.

[#21]

@fsolive post the full log in the " insert code tag "