[Telemetry] Allow custom upload bucket names

telemetry/telemetry/internal/story_runner.py

Index: telemetry/telemetry/internal/story_runner.py
diff --git a/telemetry/telemetry/internal/story_runner.py b/telemetry/telemetry/internal/story_runner.py
index cd0c9865b243e90a8b3916248d6c487ce7e780d7..5661f48087d7b2447854ce12c0e24d7a747cb6ce 100644
--- a/telemetry/telemetry/internal/story_runner.py
+++ b/telemetry/telemetry/internal/story_runner.py
@@ -321,8 +321,10 @@ def RunBenchmark(benchmark, finder_options):
       return_code = 255
 
     try:
-      bucket = cloud_storage.BUCKET_ALIASES[finder_options.upload_bucket]
       if finder_options.upload_results:
+        bucket = finder_options.upload_bucket

[Zhen Wang, 2016/08/12 17:35:27]

Since now we allow arbitrary bucket name, if the bucket name is not valid (e.g.,
typo), is there a way to do some error handling for that? For example, print out
error message to notify the user?

[perezju, 2016/08/12 17:41:02]

I thought about that. The user will get some log error messages, but right in
the end, after the benchmark has finished running; letting them know that traces
could not be uploaded.

I guess we could try to validate that the bucket is reachable and with the right
permissions right after parsing the command line options?

Will work on that, but probably wont have it till next Wednesday (as I'm flying
+ in training till then).

[Zhen Wang, 2016/08/12 17:48:21]

Where do the custom buckets come from? Is that for local testing or something
else? I am just thinking if it is valuable to know the bucket name is wrong
before running the benchmark. If not, knowing that at the end is also fine.

+        if bucket in cloud_storage.BUCKET_ALIASES:
+          bucket = cloud_storage.BUCKET_ALIASES[bucket]
         results.UploadTraceFilesToCloud(bucket)
         results.UploadProfilingFilesToCloud(bucket)
     finally: