Check Cassandra logs for tombstone threshold error On Aug 3, 2015 7:32 PM, "Robert Coli" <rc...@eventbrite.com> wrote:
> On Mon, Aug 3, 2015 at 2:48 PM, Sid Tantia <sid.tan...@baseboxsoftware.com > > wrote: > >> Any select all or select count query on a particular table is timing out >> with "Cassandra::Errors::TimeoutError: Timed out" >> >> A “SELECT * FROM <table> WHERE <partition key> = ‘<partition key value>’ >> on the table works, but a “SELECT * FROM <table> LIMIT 1; does not work. >> All other tables and queries work. >> >> Any ideas as to why this might be happening? >> > > This specific question is getting to be a FAQ... > > Briefly : > > 1) Cassandra operates best when you supply PRIMARY KEY with your request. > 2) The worst case scenario is therefore SELECT * FROM Table; > 3) While SELECT * FROM Table; is trivial in a non-distributed RDBMS, > Cassandra is ****NOT A NON-DISTRIBUTED RDBMS**** and so it is NOT TRIVIAL > IN CASSANDRA > 4) Cassandra is, instead, a distributed data-store in which most > operations default to a timeout of fewer than a single digit number of > seconds > 5) So if SELECTing * FROM your Table is likely to take longer than that, > what you'll get is a timeout > > The stock remediation for this FAQ is : > > a) don't use Cassandra for cases where you need to SELECT * FROM Table > b) if you do have to use it for such a case, use a driver with internal > pagination > > =Rob > >
