New options for managing broad crawling queue get algorithm and improved logging in manager and strategy worker.
0.5.2.3
See https://github.com/scrapinghub/frontera/issues/173
0.5.2.2
- `CONSUMER_BATCH_SIZE` is removed and two new options are introduced `SPIDER_LOG_CONSUMER_BATCH_SIZE` and `SCORING_LOG_CONSUMER_BATCH_SIZE` - Traceback is thrown into log when SIGUSR1 is received in DBW or SW. - Finishing in SW is fixed when crawling strategy reports finished.
0.5.2.1
Before that release the default compression codec was Snappy. We found out Snappy support is broken in certain Kafka versions, and issued that release. The latest version has no compression codec enabled by default, and allows to choose the compression codec with `KAFKA_CODEC_LEGACY` option.
0.5.2
From now on, strategy worker continues to operate after internal exception. There were also minor improvements.
0.5.1.1
Graphs import was removed from frontera module, therefore SQLAlchemy isn't required anymore, when it's not used.