You can delete commitlogs and start the node. But run repair on that node to sync any data mismatch.
Regards, Nitan Cell: 510 449 9629 > On Jun 25, 2019, at 4:37 AM, pwozniak <pwozn...@man.poznan.pl> wrote: > > Hi, > > I have cluster of three Cassandra (v.2.1) machines. On one of the machines > files with commit logs filled up all available disk space (50 GB). > > I haven't change 'commitlog_total_space_in_mb', so as far as I know It > shouldn't take more that 8GB of disc space. > > I also haven't found any suspicious messages in log file and our cluster was > not hammered by huge amount of requests lately. > > This machine (cassandra process) is not able to boot up now (it crashes while > replaying commit_log) > > > 1. How can I find out what happened? > > 2. Can I just delete all commit_logs, restart machine and run repair to have > consistent data? > > 3. Maybe I can delete just part of the commit_log files so Cassandra will be > able to boot and clean (flush) all commit_log files? > > > Regards, > > Pawel > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org > For additional commands, e-mail: user-h...@cassandra.apache.org >
