Hi Brian, Thanks for your response. Yes, I did look at that post. In fact after reading that post, I set the "unchecked_tombstone_compaction" to true.
For the sstable in the example below (and its neighbors), all the data for those time windows have been compacted into a single sstable, so there is no dependency or delay caused from other sstables. Thanks, Jayesh From: <brian.spind...@gmail.com> Date: Sunday, January 28, 2018 at 1:02 PM To: <user@cassandra.apache.org> Subject: Re: TWCS not deleting expired sstables I would start here: http://thelastpickle.com/blog/2016/12/08/TWCS-part1.html Specifically the “Hints and repairs” and “Timestamp overlap” sections might be of use. -B On Jan 25, 2018, at 11:05 AM, Thakrar, Jayesh <jthak...@conversantmedia.com<mailto:jthak...@conversantmedia.com>> wrote: Wondering if I can get some pointers to what's happening here and why sstables that I think should be expired are not being dropped. Here's the table's compaction property - note also set "unchecked_tombstone_compaction" to true. compaction = {'class': 'org.apache.cassandra.db.compaction.TimeWindowCompactionStrategy', 'compaction_window_size': '7', 'compaction_window_unit': 'DAYS', 'max_threshold': '4', 'min_threshold': '4', 'unchecked_tombstone_compaction': 'true'} We insert data with timestamp and TTL programmatically. Here's one set of sstable that I expect to be removed: $ ls -lt *Data.db | tail -5 -rw-r--r--. 1 vchadoop vchadoop 31245097312 Sep 20 17:16 mc-1308-big-Data.db -rw-r--r--. 1 vchadoop vchadoop 31524316252 Sep 19 14:27 mc-1187-big-Data.db -rw-r--r--. 1 vchadoop vchadoop 21405216502 Sep 18 14:14 mc-1070-big-Data.db -rw-r--r--. 1 vchadoop vchadoop 13609890747 Sep 13 20:53 mc-178-big-Data.db $ date +%s 1516895877 $ date Thu Jan 25 15:58:00 UTC 2018 $ sstablemetadata $PWD/mc-130-big-Data.db | head -20 SSTable: /ae/disk1/data/ae/raw_logs_by_user-f58b9960980311e79ac26928246f09c1/mc-130-big Partitioner: org.apache.cassandra.dht.Murmur3Partitioner Bloom Filter FP chance: 0.010000 Minimum timestamp: 1496602800000000 Maximum timestamp: 1498078800000000 SSTable min local deletion time: 1507924954 SSTable max local deletion time: 1509400832 Compressor: org.apache.cassandra.io.compress.LZ4Compressor Compression ratio: 0.17430158132352797 TTL min: 2630598 TTL max: 4086188 First token: -9177441867697829836 (key=823134638755651936) Last token: 9155171035305804798 (key=395118640769012487) minClustringValues: [-1, da, 3, 1498082382078, -9223371818124305448, -9223371652504795402, -1] maxClustringValues: [61818, tpt, 325, 1496602800000, -4611686088173246790, 9223372014135560885, 1] Estimated droppable tombstones: 1.1983492967652476 SSTable Level: 0 Repaired at: 0 Replay positions covered: {CommitLogPosition(segmentId=1505171071629, position=7157684)=CommitLogPosition(segmentId=1505171075152, position=6263269)} totalColumnsSet: 111047277
