Mercurial > p > roundup > code
comparison roundup/rest.py @ 7157:68ff3d2a9f6b
Handle KeyError if rate limit refills between update and status
| author | John Rouillard <rouilj@ieee.org> |
|---|---|
| date | Thu, 23 Feb 2023 15:43:54 -0500 |
| parents | 6f09103a6522 |
| children | 89f84f0d7cd3 |
comparison
equal
deleted
inserted
replaced
| 7156:6f09103a6522 | 7157:68ff3d2a9f6b |
|---|---|
| 2073 for header, value in limitStatus.items(): | 2073 for header, value in limitStatus.items(): |
| 2074 self.client.setHeader(header, value) | 2074 self.client.setHeader(header, value) |
| 2075 # User exceeded limits: tell humans how long to wait | 2075 # User exceeded limits: tell humans how long to wait |
| 2076 # Headers above will do the right thing for api | 2076 # Headers above will do the right thing for api |
| 2077 # aware clients. | 2077 # aware clients. |
| 2078 msg = _("Api rate limits exceeded. Please wait: %s seconds.") % limitStatus['Retry-After'] | 2078 try: |
| 2079 retry_after = limitStatus['Retry-After'] | |
| 2080 except KeyError: | |
| 2081 # handle race condition. If the time between | |
| 2082 # the call to grca.update and grca.status | |
| 2083 # is sufficient to reload the bucket by 1 | |
| 2084 # item, Retry-After will be missing from | |
| 2085 # limitStatus. So report a 1 second delay back | |
| 2086 # to the client. We treat update as sole | |
| 2087 # source of truth for exceeded rate limits. | |
| 2088 retry_after = 1 | |
| 2089 | |
| 2090 msg = _("Api rate limits exceeded. Please wait: %s seconds.") % retry_after | |
| 2079 output = self.error_obj(429, msg, source="ApiRateLimiter") | 2091 output = self.error_obj(429, msg, source="ApiRateLimiter") |
| 2080 else: | 2092 else: |
| 2081 for header, value in limitStatus.items(): | 2093 for header, value in limitStatus.items(): |
| 2082 # Retry-After will be 0 because | 2094 # Retry-After will be 0 because |
| 2083 # user still has quota available. | 2095 # user still has quota available. |