Grokbase › Groups › Hadoop › scm-users › June 2013
FAQ
Hi,
I'm trying to configure a small cluster (about 10 nodes) using cloudera
manager (last version, Cloudera Enterprise Trial 4.6.0).
I've installed the last release of CDH4 and I've integrated kerberos for
authentication.

For some reason, the manager is not able to properly communicate with
the name node.
When launching HDFS, I get
"No connection to determine the active NameNode could be made for the last
3 minute(s). Test cannot be run and will be considered failing: Test of
whether a NameNode is running, healthy and active." and the node is
reported in bad state.

However, from the command line I'm able to query the filesystem (e.g., hdfs
-dfs ls / it works perfectly)
When disabling kerberos the system goes back to normal.

It seems that some of the testing/altering controls are broken.
I do not have firewall or NAT in place.
I've also collected some tcpdump traces to check the traffic (hopefully to
see a port blocked or something) but with no luck.
The logs in /var/log/cloudera-scm-* do not not report indication of errors.

Any ideas? :(

Search Discussions

  • Gilad Wolff at Jun 20, 2013 at 1:27 pm ⇧
    Forgot to cc scm-users.

    G.
    ---------- Forwarded message ----------
    From: "Gilad Wolff" <[email protected]>
    Date: Jun 20, 2013 6:26 AM
    Subject: Re: namenode bad health when enabling kerberos with cloudera
    manager
    To: "Alessandro Finamore" <[email protected]>
    Cc:

    Hi Alessandro,

    Have you restarted the monitoring services after you enabled kerberos on
    your cluster? The easiest way is to restart all the mgmt services. If you
    have and still experience this, can you send me the service monitor log?

    G.
    On Jun 20, 2013 3:09 AM, "Alessandro Finamore" wrote:

    Hi,
    I'm trying to configure a small cluster (about 10 nodes) using cloudera
    manager (last version, Cloudera Enterprise Trial 4.6.0).
    I've installed the last release of CDH4 and I've integrated kerberos for
    authentication.

    For some reason, the manager is not able to properly communicate with
    the name node.
    When launching HDFS, I get
    "No connection to determine the active NameNode could be made for the
    last 3 minute(s). Test cannot be run and will be considered failing: Test
    of whether a NameNode is running, healthy and active." and the node is
    reported in bad state.

    However, from the command line I'm able to query the filesystem (e.g.,
    hdfs -dfs ls / it works perfectly)
    When disabling kerberos the system goes back to normal.

    It seems that some of the testing/altering controls are broken.
    I do not have firewall or NAT in place.
    I've also collected some tcpdump traces to check the traffic (hopefully to
    see a port blocked or something) but with no luck.
    The logs in /var/log/cloudera-scm-* do not not report indication of errors.

    Any ideas? :(

  • Finamore A. at Jun 20, 2013 at 3:17 pm ⇧
    Hi Gilad,
    thanks for the reply.

    You are right.
    I found the solution by myself like 2 seconds after I posted the message.
    doh! :)


    On 20 June 2013 15:27, Gilad Wolff wrote:
    Forgot to cc scm-users.

    G.

    ---------- Forwarded message ----------
    From: "Gilad Wolff" <[email protected]>
    Date: Jun 20, 2013 6:26 AM
    Subject: Re: namenode bad health when enabling kerberos with cloudera
    manager
    To: "Alessandro Finamore" <[email protected]>
    Cc:

    Hi Alessandro,

    Have you restarted the monitoring services after you enabled kerberos on
    your cluster? The easiest way is to restart all the mgmt services. If you
    have and still experience this, can you send me the service monitor log?

    G.
    On Jun 20, 2013 3:09 AM, "Alessandro Finamore" wrote:

    Hi,
    I'm trying to configure a small cluster (about 10 nodes) using cloudera
    manager (last version, Cloudera Enterprise Trial 4.6.0).
    I've installed the last release of CDH4 and I've integrated kerberos for
    authentication.

    For some reason, the manager is not able to properly communicate with the
    name node.
    When launching HDFS, I get
    "No connection to determine the active NameNode could be made for the last
    3 minute(s). Test cannot be run and will be considered failing: Test of
    whether a NameNode is running, healthy and active." and the node is reported
    in bad state.

    However, from the command line I'm able to query the filesystem (e.g.,
    hdfs -dfs ls / it works perfectly)
    When disabling kerberos the system goes back to normal.

    It seems that some of the testing/altering controls are broken.
    I do not have firewall or NAT in place.
    I've also collected some tcpdump traces to check the traffic (hopefully to
    see a port blocked or something) but with no luck.
    The logs in /var/log/cloudera-scm-* do not not report indication of
    errors.

    Any ideas? :(


    --
    --------------------------------------------------
    Alessandro Finamore, PhD
    Politecnico di Torino
    --
    Office: +39 0115644234
    Mobile: +39 3280251485
    SkypeId: alessandro.finamore
    ---------------------------------------------------

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupscm-users @
categorieshadoop
postedJun 20, '13 at 10:09a
activeJun 20, '13 at 3:17p
posts3
users3
websitecloudera.com
irc#hadoop

3 users in discussion

Finamore A.: 1 post Alessandro Finamore: 1 post Gilad Wolff: 1 post

Content

People

Support

Translate

site design / logo © 2023 Grokbase