Create a schema
This differs from bug 789493 in that we only want to store symbol *metadata* in here to make symbol space reporting and cleanup simpler than it currently is. This ties into bug 958206 because we'll need an actual symbol upload endpoint to receive symbols in order to fill the database.
We have a "symbol upload endpoint" now that passes the files uploads through (validates them and stores metadata about the upload). The destination is the S3 bucket of all symbols. Does that mean this bug is invalid now?
Component: General → Symbols
We need to talk about symbol cleanup. We should cover that in Portland. If what we currently have in the database is good enough then this can be closed.
Do we have a record of *used* symbols? In the processor we ping postgres about the missing symbol when it happens, but should we perhaps ping postgres when we do use a symbol? (by ping I mean, `update symbols set used=now() where symbol=%s`) That way we'd know which ones we almost never use.
(In reply to Ted Mielczarek [:ted.mielczarek] from comment #3) > We need to talk about symbol cleanup. We should cover that in Portland. If > what we currently have in the database is good enough then this can be > closed. The topic of cleanup was covered in Portland. We concluded that as long as we set a really long S3 life-cycle (I believe it's set to 2 years) we solve the problem of cleanup and not removing old OS based symbols. Yes, there is a small chance that someone debugs a 2+ old build and gets sadface but that's the tradeoff we decided. Additionally, with a 2 year life-cycle we calculated that the cost is not a problem to worry about. So if this bug is all about cleanup, I'm going to resolve this as invalid now.
Status: NEW → RESOLVED
Last Resolved: a year ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.