kafka-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shrikant Patel <shripatel1...@gmail.com>
Subject Trying to understand - timeindex and index
Date Fri, 12 Jul 2019 16:15:59 GMT
hi

We use kafka_2.11-2.0.0.

With our partitions i see following 4 files in each partition directory. I
have done basic research and understand what each file is used for.

       0 Jul 11 19:17 leader-epoch-checkpoint
       0 Jul 11 19:17 00000000000000000000.log
10485756 Jul 11 19:26 00000000000000000000.timeindex
10485760 Jul 11 19:26 00000000000000000000.index

This particular topic and partition is old and was not used in past few
day, our log retention is set to log.retention.hours=24 (global). So it
deleted the old message on each topic, and *.log is 0 bytes but time and
timeindex are 10MB each.* I don't see how or what cleans up the timeindex
or index files. Any insights?*

My concern we have 100 topic and each has 45 partitions, so this files
combine take up (100*45*20MB= 90GB) ~100 GB.

These 2 files have same size irrespective of *.log file. As below its 10MB
for topic that actively being used.

       10 Jul  7 15:03 00000000000003722247.snapshot
       16 Jul  8 15:05 leader-epoch-checkpoint
 10485756 Jul 12 16:03 00000000000003722247.timeindex
 10485760 Jul 12 16:03 00000000000003722247.index
317096431 Jul 12 16:03 00000000000003722247.log

Thanks
Shri

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message