Introduction to HDFS Logs

Log Description

Log path: The default path of HDFS logs is /var/log/Bigdata/hdfs/Role name.

  • NameNode: /var/log/Bigdata/hdfs/nn (run logs) and /var/log/Bigdata/audit/hdfs/nn (audit logs)

  • DataNode: /var/log/Bigdata/hdfs/dn (run logs) and /var/log/Bigdata/audit/hdfs/dn (audit logs)

  • ZKFC: /var/log/Bigdata/hdfs/zkfc (run logs) and /var/log/Bigdata/audit/hdfs/zkfc (audit logs)

  • JournalNode: /var/log/Bigdata/hdfs/jn (run logs) and /var/log/Bigdata/audit/hdfs/jn (audit logs)

  • Router: /var/log/Bigdata/hdfs/router (run logs) and /var/log/Bigdata/audit/hdfs/router (audit logs)

  • HttpFS: /var/log/Bigdata/hdfs/httpfs (run logs) and /var/log/Bigdata/audit/hdfs/httpfs (audit logs)

Log archive rule: The automatic HDFS log compression function is enabled. By default, when the size of logs exceeds 100 MB, logs are automatically compressed into a log file named in the following format: <Original log file name>-<yyyy-mm-dd_hh-mm-ss.[ID].log.zip. A maximum of 100 latest compressed files are reserved. The number of compressed files can be configured on Manager.

Table 1 HDFS log list

Type

Name

Description

Run log

hadoop-<SSH_USER>-<process_name>-<hostname>.log

HDFS system log, which records most of the logs generated when the HDFS system is running.

hadoop-<SSH_USER>-<process_name>-<hostname>.out

Log that records the HDFS running environment information.

hadoop.log

Log that records the operation of the Hadoop client.

hdfs-period-check.log

Log that records scripts that are executed periodically, including automatic balancing, data migration, and JournalNode data synchronization detection.

<process_name>-<SSH_USER>-<DATE>-<PID>-gc.log

Garbage collection log file

postinstallDetail.log

Work log before the HDFS service startup and after the installation.

hdfs-service-check.log

Log that records whether the HDFS service starts successfully.

hdfs-set-storage-policy.log

Log that records the HDFS data storage policies.

cleanupDetail.log

Log that records the cleanup logs about the uninstallation of the HDFS service.

prestartDetail.log

Log that records cluster operations before the HDFS service startup.

hdfs-recover-fsimage.log

Recovery log of the NameNode metadata.

datanode-disk-check.log

Log that records the disk status check during the cluster installation and use.

hdfs-availability-check.log

Log that check whether the HDFS service is available.

hdfs-backup-fsimage.log

Backup log of the NameNode metadata.

startDetail.log

Detailed log that records the HDFS service startup.

hdfs-blockplacement.log

Log that records the placement policy of HDFS blocks.

upgradeDetail.log

Upgrade logs.

hdfs-clean-acls-java.log

Log that records the clearing of deleted roles' ACL information by HDFS.

hdfs-haCheck.log

Run log that checks whether the NameNode in active or standby state has obtained scripts.

<process_name>-jvmpause.log

Log that records JVM pauses during process running.

hadoop-<SSH_USER>-balancer-<hostname>.log

Run log of HDFS automatic balancing.

hadoop-<SSH_USER>-balancer-<hostname>.out

Log that records information of the environment where HDFS executes automatic balancing.

hdfs-switch-namenode.log

Run log that records the HDFS active/standby switchover.

hdfs-router-admin.log

Run log of the mount table management operation

Tomcat logs

hadoop-omm-host1.out, httpfs-catalina.<DATE>.log, httpfs-host-manager.<DATE>.log, httpfs-localhost.<DATE>.log, httpfs-manager.<DATE>.log, localhost_access_web_log.log

Tomcat run log

Audit log

hdfs-audit-<process_name>.log

ranger-plugin-audit.log

Audit log that records the HDFS operations (such as creating, deleting, modifying and querying files).

SecurityAuth.audit

HDFS security audit log.

Log Level

Table 2 lists the log levels supported by HDFS. The log levels include FATAL, ERROR, WARN, INFO, and DEBUG. Logs of which the levels are higher than or equal to the set level will be printed by programs. The higher the log level is set, the fewer the logs are recorded.

Table 2 Log levels

Level

Description

FATAL

Indicates the critical error information about system running.

ERROR

Indicates the error information about system running.

WARN

Indicates that the current event processing exists exceptions.

INFO

Indicates that the system and events are running properly.

DEBUG

Indicates the system and system debugging information.

To modify log levels, perform the following operations:

  1. Go to the All Configurations page of HDFS by referring to Modifying Cluster Service Configuration Parameters.

  2. On the left menu bar, select the log menu of the target role.

  3. Select a desired log level.

  4. Save the configuration. In the displayed dialog box, click OK to make the configurations take effect.

    Note

    The configurations take effect immediately without restarting the service.

Log Formats

The following table lists the HDFS log formats.

Table 3 Log formats

Type

Format

Example

Run log

<yyyy-MM-dd HH:mm:ss,SSS>|<Log level>|<Name of the thread that generates the log>|<Message in the log>|<Location where the log event occurs>

2015-01-26 18:43:42,840 | INFO | IPC Server handler 40 on 8020 | Rolling edit logs | org.apache.hadoop.hdfs.server.namenode.FSEditLog.rollEditLog(FSEditLog.java:1096)

Audit log

<yyyy-MM-dd HH:mm:ss,SSS>|<Log level>|<Name of the thread that generates the log>|<Message in the log>|<Location where the log event occurs>

2015-01-26 18:44:42,607 | INFO | IPC Server handler 32 on 8020 | allowed=true ugi=hbase (auth:SIMPLE) ip=/10.177.112.145 cmd=getfileinfo src=/hbase/WALs/hghoulaslx410,16020,1421743096083/hghoulaslx410%2C16020%2C1421743096083.1422268722795 dst=null perm=null | org.apache.hadoop.hdfs.server.namenode.FSNamesystem$DefaultAuditLogger.logAuditMessage(FSNamesystem.java:7950)