Last modified: 2014-10-15 23:22:41 UTC
After the deployment of webstatscollector's filter on 2014-10-15, udp2log-lucene did not restart properly, and Icinga alerted about it being down. Since service status indicated other udp2log instances were down too [1] all udp2log instances on oxygen got restarted on 2014-10-15 20:22. During this restart (<1 minute), no messages have been collected. [1] Now seeing the hourly file on disk, it seems the main udp2log instance was working fine, only lucene's udp2log causing problems. So data up to the restart and after the restart is good.
Note that this only affects the files in http://dumps.wikimedia.org/other/pagecounts-raw . The files in http://dumps.wikimedia.org/other/pagecounts-all-sites are not affected.
