ALM-45645 Pending Flush Size of RocksDB Continuously Exceeds the Threshold

Alarm Description

The system checks the RocksDB monitoring data of jobs at the user-specified alarm reporting interval (metrics.reporter.alarm.job.alarm.rocksdb.metrics.duration, 180s by default). This alarm is generated when the number of pending flush requests of RocksDB for a job continuously reaches n times the number of flush/compaction threads. This alarm is cleared when the number of pending flush requests of RocksDB for the job is less than or equal to the threshold.

• The number of flush/compaction threads is the value of state.backend.rocksdb.thread.num. The default value is 2. If SPINNING_DISK_OPTIMIZED_HIGH_MEM is enabled, the default value is 4.
• The metrics.reporter.alarm.job.alarm.rocksdb.background.jobs.multiplier parameter specifies n times the number of flush/compaction threads. The default value is 2.

Alarm Attributes

Alarm Parameters

Impact on the System

The checkpoint performance of Flink jobs are affected. There is no impact on the FlinkServer.

Possible Causes

The number of pending flush requests of RocksDB for the Flink job is too large.

Handling Procedure

Check TaskManager logs for the number of pending flush requests and collect logs.

1. Log in to FusionInsight Manager as a user who has the FlinkServer management permission.
2. Choose O&M > Alarm > Alarms > ALM-45645 Pending Flush Size of RocksDB Continuously Exceeds the Threshold, view Location, and obtain the name of the task for which the alarm is generated.
3. Choose Cluster > Services > Yarn and click the link next to ResourceManager WebUI to go to the native Yarn page.

4. Locate the abnormal task based on its name displayed in Location, search for and record the application ID of the job, and check whether the job logs are available on the Yarn page.
   Figure 1 Application ID of a job
   
   • If yes, go to 5.
   • If no, go to 6.

5. Click the application ID of the failed job to go to the job page.
   1. Click Logs in the Logs column to view JobManager logs.
      Figure 2 Clicking Logs
      
   2. Click the ID in the Attempt ID column and click Logs in the Logs column to view and save TaskManager logs. Then go to 7.
      Figure 3 Clicking the ID in the Attempt ID column
      
      Figure 4 Clicking Logs
      
      

      You can also log in to Manager as a user who has the management permission for the current Flink job. Choose Cluster > Services > Flink, and click the link next to Flink WebUI. On the displayed Flink web UI, click Job Management, click More in the Operation column, and select Job Monitoring to view TaskManager logs.

If logs are unavailable on the Yarn page, download logs from HDFS.

6. On Manager, choose Cluster > Services > HDFS, click the link next to NameNode WebUI to go to the HDFS page, choose Utilities > Browse the file system, and download logs in the /tmp/logs/Username/bucket-logs-tfile/Last four digits of the task application ID/Application ID of the task directory.

Check whether there are too many pending flush requests.

7. Check whether the sum of the values of rocksdb.mem-table-flush-pending and rocksdb.compaction-pending in TaskManager monitoring logs (keyword RocksDBMetricPrint) is greater than n times the number of RocksDB threads (metrics.reporter.alarm.job.alarm.rocksdb.background.jobs.multiplier, 2 by default). If it is, you can increase the number of RocksDB threads.
   • If yes, adjust the values of the following custom parameters on the job development page of the Flink web UI, save the settings, and go to 8.

     Table 1 Custom parameters

     Parameter	Default Value	Description
     state.backend.rocksdb.thread.num	2 4: enables SPINNING_DISK_OPTIMIZED_HIGH_MEM.	Number of flush threads. Increase the number of threads to quickly flush memory data to disks. When the number of threads is increased, the number of vCores also needs to be increased. 2 to 10 are recommended.

   • If no, go to 9.

8. Restart the job and check whether the alarm is cleared.
   • If yes, no further action is required.
   • If no, go to 9.

9. Contact O&M personnel and send the collected logs.

Alarm Clearance

This alarm is automatically cleared after the fault is rectified.

Related Information

None.