once we get the symbols hosted in s3 we'll want to help migrate snppy to use the new symbol store.

Vladan: I don't know where bugs about the Snappy server normally go (GitHub?) but we're going to be moving symbol storage from the NetApp to S3 this quarter. It doesn't look like this will be really hard in the snappy server since it's written in Python. It will just involve swapping out an open() call for a urlopen() call, and dropping the symbolPaths config option in favor of a symbolURL config option.

I'll fix this, it looks easy.

I had to do some refactoring to make this work properly, but I think the end result is good.

Comment on attachment 8550962 [details] Add support for fetching symbols over HTTP Merged your pull request, just make the changes i suggested in another pull request

Did you test with both file paths & urls?

Yes, I tested with a local Windows build, using a local path for its symbols and falling back to HTTP for system symbols and it worked fine. I also did some ad-hoc testing with curl and everything seems to work properly. Follow-up to address those review comments and fix one thing I thought of after the fact: https://github.com/vdjeric/Snappy-Symbolication-Server/pull/20

rhelmer: this is done, right?

Yep, thanks for following up!