mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-07-20 21:48:52 -04:00
chore: formatting
This commit is contained in:
parent
18a976f82c
commit
0dca9f3861
2 changed files with 13 additions and 4 deletions
|
@ -224,7 +224,11 @@ class RethinkDbFrontier:
|
||||||
index="priority_by_site",
|
index="priority_by_site",
|
||||||
)
|
)
|
||||||
.order_by(index=r.desc("priority_by_site"))
|
.order_by(index=r.desc("priority_by_site"))
|
||||||
.filter(lambda page: r.or_(page.has_fields("retry_after").not_(), r.now() > page["retry_after"]))
|
.filter(
|
||||||
|
lambda page: r.or_(
|
||||||
|
page.has_fields("retry_after").not_(), r.now() > page["retry_after"]
|
||||||
|
)
|
||||||
|
)
|
||||||
.limit(1)
|
.limit(1)
|
||||||
.update(
|
.update(
|
||||||
{"claimed": True, "last_claimed_by": worker_id}, return_changes="always"
|
{"claimed": True, "last_claimed_by": worker_id}, return_changes="always"
|
||||||
|
|
|
@ -505,15 +505,20 @@ class BrozzlerWorker:
|
||||||
except (brozzler.PageConnectionError, Exception) as e:
|
except (brozzler.PageConnectionError, Exception) as e:
|
||||||
if isinstance(e, brozzler.PageConnectionError):
|
if isinstance(e, brozzler.PageConnectionError):
|
||||||
self.logger.error(
|
self.logger.error(
|
||||||
"Page status code possibly indicates connection failure between host and warcprox: site=%r page=%r", site, page, exc_info=True
|
"Page status code possibly indicates connection failure between host and warcprox: site=%r page=%r",
|
||||||
|
site,
|
||||||
|
page,
|
||||||
|
exc_info=True,
|
||||||
)
|
)
|
||||||
else:
|
else:
|
||||||
self.logger.error(
|
self.logger.error(
|
||||||
"unexpected exception site=%r page=%r", site, page, exc_info=True
|
"unexpected exception site=%r page=%r", site, page, exc_info=True
|
||||||
)
|
)
|
||||||
if page:
|
if page:
|
||||||
retry_delay = min(60, 60 * (1.5 ** page.failed_attempts))
|
retry_delay = min(60, 60 * (1.5**page.failed_attempts))
|
||||||
page.retry_after = doublethink.utcnow() + datetime.timedelta(seconds=retry_delay)
|
page.retry_after = doublethink.utcnow() + datetime.timedelta(
|
||||||
|
seconds=retry_delay
|
||||||
|
)
|
||||||
page.failed_attempts = (page.failed_attempts or 0) + 1
|
page.failed_attempts = (page.failed_attempts or 0) + 1
|
||||||
if page.failed_attempts >= brozzler.MAX_PAGE_FAILURES:
|
if page.failed_attempts >= brozzler.MAX_PAGE_FAILURES:
|
||||||
self.logger.info(
|
self.logger.info(
|
||||||
|
|
Loading…
Add table
Add a link
Reference in a new issue