You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Interrupt instead of fail crawl when not fatal (#973)
- Add logger.interrupt() which results in crawler being restarted, but
without failing the crawl, unlike logger.fatal()
- Switch a few logger.fatal() -> logger.interrupt(), when crawl could be
retried, only use fatal() when initial crawl config is invalid for the
most part
- Don't increment 'failing' and set to 'failed' after unexpected errors,
set to interrupted as well to allow for retry (and exponential backoff
if in k8s)
- Allow logger to manage opening/closing its open external log file, via
setOutputFile() and closeLog()
- Move closeLog() and setStatusAndExit() to logger itself so
logger.interrupt() and logger.fatal() can call these
- Use the newer logger.interrupt() for: proxy connection unavailable,
out of disk space, interrupt signal and WACZ upload failed
---------
Co-authored-by: Tessa Walsh <tessa@bitarchivist.net>
describe: "Comma-separated list of log levels to include in logs",
283
+
describe:
284
+
"Comma-separated list of log levels to include in logs. By default all but debug are included. To include debug messages in logs, debug must also be passed to logging option.",
0 commit comments