FAQ
Is there a way to distributed an activated parcels to a host in cluster?
During the first time I down load that parcel, I didn't choose all hosts in
cluster to distribute.

I'm running a Cloudera Standard 4.7.2.

Thanks,

To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.

Search Discussions

  • Philip Langdale at Oct 24, 2013 at 8:17 pm
    Hi,

    Distribution is always done for an entire logical cluster in CM. You don't
    choose individual hosts when you distribute. If you add a new host to a
    cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in cluster?
    During the first time I down load that parcel, I didn't choose all hosts
    in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it, send an
    email to scm-users+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Philip Langdale at Oct 24, 2013 at 9:24 pm
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the impala
    parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel, except
    the one which is using cloudera manager. I don't know how it can in this
    case. Missing that parcel(impala) cause alertpublisher failed to start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in cluster?
    During the first time I down load that parcel, I didn't choose all hosts
    in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Oct 25, 2013 at 12:31 am
    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the impala
    parcel distributed and activated on the cluster?

    --phil


    On 24 October 2013 13:59, Soundwave Li <liha...@gmail.com <javascript:>>wrote:
    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Philip Langdale at Oct 31, 2013 at 4:33 pm
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has the
    Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the impala
    parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels_available(self.raw['**parcels'])
    File "/usr/lib/cmf/agent/src/cmf/**parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("**Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,**Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Nov 4, 2013 at 7:17 pm
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include service
    monitor, activity monitor, host monitor event server and alert publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH -> CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with I
    have not tried to restart them. When I tried to restart alert publisher. I
    got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
         new_process.activate()
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
         self.agent.repo.ensure_parcels_available(self.raw['parcels'])
       File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
         raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has the
    Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil


    On 24 October 2013 17:31, Soundwave Li <liha...@gmail.com <javascript:>>wrote:
    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the impala
    parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels_available(self.raw['**parcels'])
    File "/usr/lib/cmf/agent/src/cmf/**parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("**Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,**Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+...@cloudera.org <javascript:>.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Nov 4, 2013 at 7:34 pm
    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include service
    monitor, activity monitor, host monitor event server and alert publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH -> CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with I
    have not tried to restart them. When I tried to restart alert publisher. I
    got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has the
    Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels_available(self.raw['**parcels'])
    File "/usr/lib/cmf/agent/src/cmf/**parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("**Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,**Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Philip Langdale at Nov 4, 2013 at 8:44 pm
    You should take a look inside the IMPALA parcel directory on that machine.
    It may not actually be valid and complete. Have you run out of disk space?
    There's improved error reporting in the CM5 beta that will report if this
    happened, but in older releases, you won't see it outside the agent's log.

    --phil

    On 4 November 2013 11:34, Soundwave Li wrote:

    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include service
    monitor, activity monitor, host monitor event server and alert publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH -> CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with I
    have not tried to restart them. When I tried to restart alert publisher. I
    got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has the
    Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM. You
    don't choose individual hosts when you distribute. If you add a new host to
    a cluster, then CM will automatically distribute the parcels - you should
    never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Nov 4, 2013 at 9:58 pm
    The space is enough and the size of this folder looks ok. Is there any way
    to check it this parcel is valid and complete?

    582M ./IMPALA-1.1.1-1.p0.17/

    Thanks,
    Hang

    在 2013年11月4日星期一UTC-8下午12时44分43秒,Philip Langdale写道:
    You should take a look inside the IMPALA parcel directory on that machine.
    It may not actually be valid and complete. Have you run out of disk space?
    There's improved error reporting in the CM5 beta that will report if this
    happened, but in older releases, you won't see it outside the agent's log.

    --phil


    On 4 November 2013 11:34, Soundwave Li <liha...@gmail.com <javascript:>>wrote:
    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include
    service monitor, activity monitor, host monitor event server and alert
    publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH ->
    CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with I
    have not tried to restart them. When I tried to restart alert publisher. I
    got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-**ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**agent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels_available(self.raw['**parcels'])
    File "/usr/lib/cmf/agent/src/cmf/**parcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("**Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,**Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has the
    Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,**Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/**ag**ent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**ag**ent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels**_available(self.raw['**parcels']**)
    File "/usr/lib/cmf/agent/src/cmf/**pa**rcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("**P**arcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,****Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM.
    You don't choose individual hosts when you distribute. If you add a new
    host to a cluster, then CM will automatically distribute the parcels - you
    should never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose all
    hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org <javascript:>.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Philip Langdale at Nov 5, 2013 at 12:14 am
    Look in the agent log for that machine. It will show it enumerating the
    available parcels. If it tries to load the impala one and fails, you'll see
    an error

    --phil

    On 4 November 2013 13:57, Soundwave Li wrote:

    The space is enough and the size of this folder looks ok. Is there any way
    to check it this parcel is valid and complete?

    582M ./IMPALA-1.1.1-1.p0.17/

    Thanks,
    Hang

    在 2013年11月4日星期一UTC-8下午12时44分43秒,Philip Langdale写道:
    You should take a look inside the IMPALA parcel directory on that
    machine. It may not actually be valid and complete. Have you run out of
    disk space? There's improved error reporting in the CM5 beta that will
    report if this happened, but in older releases, you won't see it outside
    the agent's log.

    --phil

    On 4 November 2013 11:34, Soundwave Li wrote:

    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include
    service monitor, activity monitor, host monitor event server and alert
    publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH ->
    CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with I
    have not tried to restart them. When I tried to restart alert publisher. I
    got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has
    the Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that parcel,
    except the one which is using cloudera manager. I don't know how it can in
    this case. Missing that parcel(impala) cause alertpublisher failed to
    start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM.
    You don't choose individual hosts when you distribute. If you add a new
    host to a cluster, then CM will automatically distribute the parcels - you
    should never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host in
    cluster?
    During the first time I down load that parcel, I didn't choose
    all hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Nov 5, 2013 at 12:51 am
    I feel impala parcel is not active correctly or has some other issue.
    I find a lot of similar exceptions in agent log.

      user': u'root', u'auto_restart': False, u'run_generation': 1, u'parcels':
    {u'CDH': u'4.4.0-1.cdh4.4.0.p0.39', u'IMPALA': u'1.1.1-1.p0.17'},
    u'environment': {}, u'optional_tags': [], u'running': True, u'program':
    u'mgmt/mgmt.sh', u'required_tags': [], u'arguments': [u'inspector',
    u'input.json', u'output.json', u'DEFAULT'], u'special_file_info': [],
    u'group': u'root', u'id': 396, u'resources': [], u'one_off': True}
    Traceback (most recent call last):
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in
    handle_heartbeat_response
         new_process.activate()
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
         self.agent.repo.ensure_parcels_available(self.raw['parcels'])
       File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in
    ensure_parcels_available
         raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.

    Does it mean if I can reinstall impala parcel on this host, this exception
    will be fixed?

    -H

    在 2013年11月4日星期一UTC-8下午4时14分01秒,Philip Langdale写道:
    Look in the agent log for that machine. It will show it enumerating the
    available parcels. If it tries to load the impala one and fails, you'll see
    an error

    --phil


    On 4 November 2013 13:57, Soundwave Li <liha...@gmail.com <javascript:>>wrote:
    The space is enough and the size of this folder looks ok. Is there any
    way to check it this parcel is valid and complete?

    582M ./IMPALA-1.1.1-1.p0.17/

    Thanks,
    Hang

    在 2013年11月4日星期一UTC-8下午12时44分43秒,Philip Langdale写道:
    You should take a look inside the IMPALA parcel directory on that
    machine. It may not actually be valid and complete. Have you run out of
    disk space? There's improved error reporting in the CM5 beta that will
    report if this happened, but in older releases, you won't see it outside
    the agent's log.

    --phil

    On 4 November 2013 11:34, Soundwave Li wrote:

    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,**Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include
    service monitor, activity monitor, host monitor event server and alert
    publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH ->
    CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with
    I have not tried to restart them. When I tried to restart alert publisher.
    I got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-**ALERTPUBLISHER**', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/**ag**ent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**ag**ent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels**_available(self.raw['**parcels']**)
    File "/usr/lib/cmf/agent/src/cmf/**pa**rcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("**P**arcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,****Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has
    the Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,****Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that
    parcel, except the one which is using cloudera manager. I don't know how it
    can in this case. Missing that parcel(impala) cause alertpublisher failed
    to start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/**ag****ent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/**ag****ent.py", line 1922, in activate
    self.agent.repo.ensure_**parcels****_available(self.raw['**parcels']****)
    File "/usr/lib/cmf/agent/src/cmf/**pa****rcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("**P****arcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,******Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM.
    You don't choose individual hosts when you distribute. If you add a new
    host to a cluster, then CM will automatically distribute the parcels - you
    should never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host
    in cluster?
    During the first time I down load that parcel, I didn't choose
    all hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from
    it, send an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@**cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+...@cloudera.org <javascript:>.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Philip Langdale at Nov 5, 2013 at 5:32 pm
    Can you attach the whole log? There will be an earlier error.

    --phil

    On 4 November 2013 16:51, Soundwave Li wrote:

    I feel impala parcel is not active correctly or has some other issue.
    I find a lot of similar exceptions in agent log.

    user': u'root', u'auto_restart': False, u'run_generation': 1, u'parcels':
    {u'CDH': u'4.4.0-1.cdh4.4.0.p0.39', u'IMPALA': u'1.1.1-1.p0.17'},
    u'environment': {}, u'optional_tags': [], u'running': True, u'program':
    u'mgmt/mgmt.sh', u'required_tags': [], u'arguments': [u'inspector',
    u'input.json', u'output.json', u'DEFAULT'], u'special_file_info': [],
    u'group': u'root', u'id': 396, u'resources': [], u'one_off': True}
    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in
    handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in
    ensure_parcels_available
    raise ParcelNotAvailableException("Parcel [%s] isn't available." %
    name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.

    Does it mean if I can reinstall impala parcel on this host, this exception
    will be fixed?

    -H

    在 2013年11月4日星期一UTC-8下午4时14分01秒,Philip Langdale写道:
    Look in the agent log for that machine. It will show it enumerating the
    available parcels. If it tries to load the impala one and fails, you'll see
    an error

    --phil

    On 4 November 2013 13:57, Soundwave Li wrote:

    The space is enough and the size of this folder looks ok. Is there any
    way to check it this parcel is valid and complete?

    582M ./IMPALA-1.1.1-1.p0.17/

    Thanks,
    Hang

    在 2013年11月4日星期一UTC-8下午12时44分43秒,Philip Langdale写道:
    You should take a look inside the IMPALA parcel directory on that
    machine. It may not actually be valid and complete. Have you run out of
    disk space? There's improved error reporting in the CM5 beta that will
    report if this happened, but in older releases, you won't see it outside
    the agent's log.

    --phil

    On 4 November 2013 11:34, Soundwave Li wrote:

    A quick update.

    Looks like manually add a symlink for IMPALA (e.g. "sudo ln -s
    IMPALA-1.1.1-1.p0.17 IMPALA") does not work...



    在 2013年11月4日星期一UTC-8上午11时17分42秒,Soundwave Li写道:
    Sorry for making you confused. Let me verify and check.

    In my cluster there are around 20 hosts.
    One host A hosts all cloudera manager related service. It include
    service monitor, activity monitor, host monitor event server and alert
    publisher.
    In host A's file system, it do have impala parcel on it.

    drwxr-xr-x 7 root root 4096 Aug 23 18:24 IMPALA-1.1.1-1.p0.17
    drwxr-xr-x 7 root root 4096 Sep 5 17:00 CDH-4.4.0-1.cdh4.4.0.p0.39
    lrwxrwxrwx 1 root root 26 Oct 4 15:03 CDH ->
    CDH-4.4.0-1.cdh4.4.0.p0.39

    But in cm host components page, impala is not there.

    All other management service works fine right now. Maybe related with
    I have not tried to restart them. When I tried to restart alert publisher.
    I got the following exception.

    Failed to activate {u'status_links': {}, u'name': u'cloudera-mgmt-ALERTPUBLISHER', u'config_generation': 3, ...

    ...

    Traceback (most recent call last):
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.


    Thanks,

    Hang


    在 2013年10月31日星期四UTC-7上午9时33分25秒,Philip Langdale写道:
    So I'm quite confused at this point.

    The host running the alert publisher is part of the cluster and has
    the Impala parcel on it? Did you manually verify that on the filesystem
    (/opt/cloudera/parcels)? Are any of the other management services working
    or not working? Are they all on the same host?

    --phil

    On 24 October 2013 17:31, Soundwave Li wrote:

    Yes, that host is part of cluster. Yes, that impala parcel has been
    distributed and actived on the cluster.
    Actually I have created an impala service which works fine.

    在 2013年10月24日星期四UTC-7下午2时24分18秒,Philip Langdale写道:
    Please keep the mailing list cc:ed.

    Is the CM host part of any cluster? It doesn't have to be. Is the
    impala parcel distributed and activated on the cluster?

    --phil

    On 24 October 2013 13:59, Soundwave Li wrote:

    Hi, Philip

    Thanks for your answer.
    You are right. Looks like most boxes of my cluster has that
    parcel, except the one which is using cloudera manager. I don't know how it
    can in this case. Missing that parcel(impala) cause alertpublisher failed
    to start.

    Exception looks like

    "Traceback (most recent call last):

    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
    new_process.activate()
    File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
    self.agent.repo.ensure_parcels_available(self.raw['parcels'])
    File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available

    raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)

    ParcelNotAvailableException: Parcel [IMPALA] isn't available."


    Any suggestion?


    Thanks,



    在 2013年10月24日星期四UTC-7下午1时17分05秒,Philip Langdale写道:
    Hi,

    Distribution is always done for an entire logical cluster in CM.
    You don't choose individual hosts when you distribute. If you add a new
    host to a cluster, then CM will automatically distribute the parcels - you
    should never need to do manual distributions to additional hosts.

    --phil

    On 24 October 2013 11:42, Soundwave Li wrote:

    Is there a way to distributed an activated parcels to a host
    in cluster?
    During the first time I down load that parcel, I didn't choose
    all hosts in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,

    To unsubscribe from this group and stop receiving emails from
    it, send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from
    it, send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it,
    send an email to scm-users+...@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send
    an email to scm-users+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.
  • Soundwave Li at Oct 24, 2013 at 9:06 pm
    Hi, Philip

    Thanks for your answer. (I accidentally reply to your just now.)
    You are correct. That parcel is in most boxes of cluster, except the one
    which is running cloudera manager.
    It cause alertpublisher failed during start.

    exception looks like this.
    "

    Traceback (most recent call last):
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1004, in handle_heartbeat_response
         new_process.activate()
       File "/usr/lib/cmf/agent/src/cmf/agent.py", line 1922, in activate
         self.agent.repo.ensure_parcels_available(self.raw['parcels'])
       File "/usr/lib/cmf/agent/src/cmf/parcel.py", line 575, in ensure_parcels_available
         raise ParcelNotAvailableException("Parcel [%s] isn't available." % name)
    ParcelNotAvailableException: Parcel [IMPALA] isn't available.

    "
    Any suggestion?
    Thanks,


    在 2013年10月24日星期四UTC-7上午11时42分47秒,Soundwave Li写道:
    Is there a way to distributed an activated parcels to a host in cluster?
    During the first time I down load that parcel, I didn't choose all hosts
    in cluster to distribute.

    I'm running a Cloudera Standard 4.7.2.

    Thanks,
    To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+unsubscribe@cloudera.org.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupscm-users @
categorieshadoop
postedOct 24, '13 at 6:42p
activeNov 5, '13 at 5:32p
posts13
users2
websitecloudera.com
irc#hadoop

2 users in discussion

Soundwave Li: 7 posts Philip Langdale: 6 posts

People

Translate

site design / logo © 2022 Grokbase