It would be useful to have a log of the timeseriesdb operations. For example to know when and by whom a particular series was read.
I was thinking of a table with the following columns:
- ts_key: Key of the series affected
- username: Who executed the operation. By default the DB user, but it should be possible to set a different username.
- tstamp: Timestamp of the operation
- operation: ID of operation. E.g. read, write, metadata_read, metadata_write, change_access, delete etc.
Such a table would grow quickly. For example with ~512 bytes per record and a million operations per day, to roughly 187GB within a year. Hence logging should be optional.
What do you think? Any obvious issues?
It would be useful to have a log of the timeseriesdb operations. For example to know when and by whom a particular series was read.
I was thinking of a table with the following columns:
Such a table would grow quickly. For example with ~512 bytes per record and a million operations per day, to roughly 187GB within a year. Hence logging should be optional.
What do you think? Any obvious issues?