After the memory for high kudu tserver will reject write requests, log is as follows:
19/06/01 13:34:12 INFO AsyncKuduClient: Invalidating location 34b1c13d04664cc8bae6689d39b08b77($kudu_tserver:7050) for tablet 858c055c456549569af77d14eaf997e5: Service unavailable: Soft memory limit exceeded (at 92.36% of capacity). See https://kudu.apache.org/releases/1.7.0-cdh5.16.1/docs/troubleshooting.html
1 View tserver detailed memory footprint
http: // kudu_tserver $ 8050 / mem-trackers
Page is divided into two parts:
Process memory usage for the overall occupancy
Memory usage by subsystem is detailed accounting, which at first glance more complicated, in fact, is a tree structure to cascade through the parent,
For example, root is fully occupied as 1.6G, root below logcache takes up 277K, and then look down
block_cache-sharded_lru_cache occupied 246M, server occupation 1.4G, server below is divided into a number of tablet, in addition to the tablet, as well as
log_block_manager occupied 438M;
If you find some tablet take up more memory, which want to see in the next table, you can adopt the following commands:
View all 2 tablet on the tserver
sudo -u should be remote_replica list $ kudu_tserver
such as:
Tablet id: e064f41775084680ab269b3cb3c21c76 State: RUNNING Table name: impala::test_db.test_table1 Partition: RANGE (dt) PARTITION 20190531 <= VALUES < 20190532 Estimated on disk size: 30.71M Schema: Schema [ 0:co1[int32 NOT NULL], 1:co2[string NOT NULL], 2:co3[string NOT NULL], 3:co4[string NOT NULL], 4:co5[string NULLABLE] ] ...
关于kudu的内存估算,和数据量大小以及tablet多少以及活跃replica多少都有关系,详见:
https://kudu.apache.org/docs/scaling_guide.html#memory