Load uid very slowly

[zhoujian1996]

Hi, everyOne，My project use flink handle too many kafka record. Every time, we restart the flink job, it may load too many uid from the hbase,so it makes consume the records very slowly。 SomeBody may give us some advice faster the restart process? Thanks!

[mindflying]

写入到hbase的时候，从name_cache或者id_cache dump到本地文件
重启flink的时候，直接从文件加载撒

[zhoujian1996]

好的，我们按照这种方式试下。有个问题请教下：opentsdb好像无法感知到hbase端regionServer的变化导致一直向错误的regionServer导入大量流量，这种你们有遇见过吗？