Details
-
Pain point
-
Resolution: Fixed
-
Neutral
-
None
-
None
-
None
Description
In recent months, the cheap AWS instance we use has died on us twice, losing all not backed up data in the process. Yesterday had and I had a meeting in which we identified a couple of short/mid-term solutions. I am documenting them here in the case the issue happens again in the future:
- m3 single data node

- t2 master + t2 data
- t2 master + m3 data
- t2 master + 2 cheap data nodes
I suggest to go with #1 for now and keep an eye on how it evolves. I will also try setting up CloudWatch emails if simple enough.
I am also including the notes over which we iterated during the meeting. Other solutions were considered, but they were discarded because too expensive changes.