You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Dmitry Sivachenko <tr...@gmail.com> on 2015/02/13 15:54:14 UTC
Log Aggregation
Hello!
I am using hadoop-2.4.1 in distributed mode. After a job completes, logs are aggregated to hdfs and are available via history server.
Sometimes logs appear very fast after the job completes (or fails), but sometimes it takes long (10-20-30 minutes).
During that period history server reports:
Logs not available for attempt_1422914757889_1881_m_000000_0. Aggregation may not be complete, Check back later or try the nodemanager at <host>
It seems that it does not depend on log size, it is not so big to take 20 minutes to copy to hdfs.
Why this can happen? How can I debug the issue to understand what is happening during that period before logs appear at history server?
Thanks!
Re: Log Aggregation
Posted by Xuan Gong <xg...@hortonworks.com>.
Hey, could you check whether the logs appear in HDFS ? If not, could you
check the NodeManager logs to find out when/whether the log aggregation
for that applications started ?
Thanks
Xuan Gong
On 2/13/15, 6:54 AM, "Dmitry Sivachenko" <tr...@gmail.com> wrote:
>Hello!
>
>I am using hadoop-2.4.1 in distributed mode. After a job completes, logs
>are aggregated to hdfs and are available via history server.
>
>Sometimes logs appear very fast after the job completes (or fails), but
>sometimes it takes long (10-20-30 minutes).
>
>During that period history server reports:
>Logs not available for attempt_1422914757889_1881_m_000000_0. Aggregation
>may not be complete, Check back later or try the nodemanager at <host>
>
>It seems that it does not depend on log size, it is not so big to take 20
>minutes to copy to hdfs.
>
>Why this can happen? How can I debug the issue to understand what is
>happening during that period before logs appear at history server?
>
>Thanks!
Re: Log Aggregation
Posted by Xuan Gong <xg...@hortonworks.com>.
Hey, could you check whether the logs appear in HDFS ? If not, could you
check the NodeManager logs to find out when/whether the log aggregation
for that applications started ?
Thanks
Xuan Gong
On 2/13/15, 6:54 AM, "Dmitry Sivachenko" <tr...@gmail.com> wrote:
>Hello!
>
>I am using hadoop-2.4.1 in distributed mode. After a job completes, logs
>are aggregated to hdfs and are available via history server.
>
>Sometimes logs appear very fast after the job completes (or fails), but
>sometimes it takes long (10-20-30 minutes).
>
>During that period history server reports:
>Logs not available for attempt_1422914757889_1881_m_000000_0. Aggregation
>may not be complete, Check back later or try the nodemanager at <host>
>
>It seems that it does not depend on log size, it is not so big to take 20
>minutes to copy to hdfs.
>
>Why this can happen? How can I debug the issue to understand what is
>happening during that period before logs appear at history server?
>
>Thanks!
Re: Log Aggregation
Posted by Xuan Gong <xg...@hortonworks.com>.
Hey, could you check whether the logs appear in HDFS ? If not, could you
check the NodeManager logs to find out when/whether the log aggregation
for that applications started ?
Thanks
Xuan Gong
On 2/13/15, 6:54 AM, "Dmitry Sivachenko" <tr...@gmail.com> wrote:
>Hello!
>
>I am using hadoop-2.4.1 in distributed mode. After a job completes, logs
>are aggregated to hdfs and are available via history server.
>
>Sometimes logs appear very fast after the job completes (or fails), but
>sometimes it takes long (10-20-30 minutes).
>
>During that period history server reports:
>Logs not available for attempt_1422914757889_1881_m_000000_0. Aggregation
>may not be complete, Check back later or try the nodemanager at <host>
>
>It seems that it does not depend on log size, it is not so big to take 20
>minutes to copy to hdfs.
>
>Why this can happen? How can I debug the issue to understand what is
>happening during that period before logs appear at history server?
>
>Thanks!
Re: Log Aggregation
Posted by Xuan Gong <xg...@hortonworks.com>.
Hey, could you check whether the logs appear in HDFS ? If not, could you
check the NodeManager logs to find out when/whether the log aggregation
for that applications started ?
Thanks
Xuan Gong
On 2/13/15, 6:54 AM, "Dmitry Sivachenko" <tr...@gmail.com> wrote:
>Hello!
>
>I am using hadoop-2.4.1 in distributed mode. After a job completes, logs
>are aggregated to hdfs and are available via history server.
>
>Sometimes logs appear very fast after the job completes (or fails), but
>sometimes it takes long (10-20-30 minutes).
>
>During that period history server reports:
>Logs not available for attempt_1422914757889_1881_m_000000_0. Aggregation
>may not be complete, Check back later or try the nodemanager at <host>
>
>It seems that it does not depend on log size, it is not so big to take 20
>minutes to copy to hdfs.
>
>Why this can happen? How can I debug the issue to understand what is
>happening during that period before logs appear at history server?
>
>Thanks!