diff options
author | Bryan Newbold <bnewbold@archive.org> | 2021-09-30 17:17:30 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2021-09-30 17:17:30 -0700 |
commit | 8e26ab264190b998e9035f0883f00340ca220822 (patch) | |
tree | 89ad479146b482e798b0adfd533177926345dc3a /python/sandcrawler | |
parent | f6125848f627ae9bfd3a36d807d2349e1c66bfe3 (diff) | |
download | sandcrawler-8e26ab264190b998e9035f0883f00340ca220822.tar.gz sandcrawler-8e26ab264190b998e9035f0883f00340ca220822.zip |
tune SPN CDX retry/wait depending on mode (priority vs daily)
Diffstat (limited to 'python/sandcrawler')
-rw-r--r-- | python/sandcrawler/ia.py | 6 | ||||
-rw-r--r-- | python/sandcrawler/ingest.py | 2 |
2 files changed, 5 insertions, 3 deletions
diff --git a/python/sandcrawler/ia.py b/python/sandcrawler/ia.py index 68b3466..a5d19cd 100644 --- a/python/sandcrawler/ia.py +++ b/python/sandcrawler/ia.py @@ -805,6 +805,8 @@ class SavePageNowClient: self.poll_count = 60 self.poll_seconds = 3.0 + self.spn_cdx_retry_sec = kwargs.get('spn_cdx_retry_sec', 9.0) + def save_url_now_v2(self, request_url, force_simple_get=0, capture_outlinks=0): """ Returns a "SavePageNowResult" (namedtuple) if SPN request was processed @@ -1023,7 +1025,7 @@ class SavePageNowClient: url=spn_result.terminal_url, datetime=spn_result.terminal_dt, filter_status_code=filter_status_code, - retry_sleep=9.0, + retry_sleep=self.spn_cdx_retry_sec, ) # sometimes there are fuzzy http/https self-redirects with the # same SURT; try to work around that @@ -1032,7 +1034,7 @@ class SavePageNowClient: url=spn_result.terminal_url, datetime=spn_result.terminal_dt, filter_status_code=200, - retry_sleep=9.0, + retry_sleep=self.spn_cdx_retry_sec, ) except KeyError as ke: print(" CDX KeyError: {}".format(ke), file=sys.stderr) diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index 1d33b94..ba478ea 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -64,7 +64,7 @@ class IngestFileWorker(SandcrawlerWorker): self.wayback_client = WaybackClient() self.spn_client = kwargs.get('spn_client') if not self.spn_client: - self.spn_client = SavePageNowClient() + self.spn_client = SavePageNowClient(kwargs.get('spn_cdx_retry_sec', 9.0)) self.grobid_client = kwargs.get('grobid_client') if not self.grobid_client: self.grobid_client = GrobidClient() |