Hi everyone! I asked the question in the "general...
# random
h
Hi everyone! I asked the question in the "general"-channel by mistake. So maybe it is besser suited here: Is it possible, to have an OFFLINE table filled by a RealtimeToOfflineSegmentsTask to have one row per primary key (e.g. hostname)? For example: everytime, a new host entry is coming in, the old row (all columns) is completely replaced? Right now, a new row is appended to the table depending on the roundBucketTimePeriod, for example "1d" generates a new row every day. The OFFLINE table is tagged as isDimTable and the REALTIME mergeType set to "dedup". Bascially, I only want the newest entry to be part of the OFFLINE-table. How do ein accomplish this?
r
I think we should enable upsert in the realtime table.
h
Thank you for the answer. Upsert only works on Realtime tables, so if data is ingested into the offline counterpart (via RealtimeToOfflineSegemntsTask), I still would have multiple entries in the offline table, would'nt I?
r
Yeah true. Any specific reason why we can’t have only a realtime table in this case?
h
Hmm, I guess nothing. I thought, since this is only an attributes table, it should be tagged-dimensional, which only works for offline mode.
r
Are you planning to execute look-up queries on top of this table? If not, we don’t need to tag it as a dimensional table.
h
You are absolutely right. Thx!
🍷 1