Logs can be challenging to interpret with large numbers of autoscaled Spanner instances
henrybell opened this issue · 0 comments
henrybell commented
When the number of target Spanner instances being autoscaled grows large, it can be difficult to interpret the logs from the poller and scaler components. Possibilities for addressing this include some combination of:
- Adding a logline prefix and/or other identifying field
- Moving to more structured logging
- Using Log Analytics to enable structured querying
- Creating a custom dashboard to be able to drill down into logs