Cassandra Tombstones related questions

[#1]

Hi Team,

A customer using BaaS Private Cloud has the Error messages of:

ERROR [ReadStage:29] 2017-07-14 03:00:39,232 SliceQueryFilter.java (line 206) Scanned over 100000 tombstones in Usergrid_Applications.Entity_Index_Entries; query aborted (see tombstone_failure_threshold)

and according to the existing cases by the other customers with the same error, they are advised to configure;

• modify the tombstone_failure_threshold: from the default 100000 to 1000000 and tombstone_warn_threshold from the default 1000 to 10000.
• modify the gc_grace period from the default 10 days to 2 days.

Then the questions from the customer are the followings:

1. How to confirm that the deletion mechanism of tombstone occurring every gc_grace_seconds is working normally
2. Which keyespace and table should be compaction targeted?
   Is this the above ‘Usergrid_Applications’ Keyspace and ‘Entity_Index_Entries’ Column Family only or the others also?
3. How to find the number of tombstones?

Could you please advise us on these points?

Regards,
Toshi

[spettypiece]

There may be a problem with changing gc_grace from the default 10 days to 2 days, since recommended once per week full 'Repair' is executed, could cause a data integrity issue.