I am probably missing something. Is there anyone who is using the commitlog to make point in time restoring ?
Is there a procedure to make it works ?
This is my procedure.
1- Add this line to cassandra-env.sh JVM_OPTS="$JVM_OPTS -Dcassandra.replayList=pns_nonreg_bench.cf1"
2- I edit the file /etc/cassandra/commitlog_archiving.properties like this
archive_command=/bin/ln %path /disk2/cassandra/commit_log_backup/%name
restore_command=cp -f %from %to
3- I restart all the nodes
4- I made a delete on cassandra and the modify the restore_point_in_time to set it before the delete
restore_point_in_time=2018:07:30 09:54:00 # before the delete
5- then I restart cassandra and I just see in the logs this
INFO [main] 2018-07-30 10:40:43,652 CommitLog.java:157 - Replaying /var/opt/hosting/db/cassandra/commitlog/CommitLog-6-1532697162576.log
INFO [main] 2018-07-30 10:40:43,705 CommitLog.java:159 - Log replay complete, 0 replayed mutations
I did not get the data back, (re write it ).
For infor I made the procedure using snapshots and the replaying the commitlogs without success
Thank you very much
"The best way to predict the future is to invent it" Alan Kay