03 Oct 23: One server has been relocated, the server currently doesn't have a public IP address in the meantime. I wrote some additional tunnel code running separately to handle this. When the server is engaged the connection is 5600+ days uniform (since around 24 July 2008). In the unlikely case I disengaged the server because of some problem the retention is 1200-3800 days depending on newsgroup. If you experience any issue please let me know.
29 Nov 23: Because of encrypted and "obfuscated" flood mass posted by few nzb websites and newzbin-like communities using usenet servers as private storage for their members to download the posts - the situation with the content is pretty chaotic, since posts are disguised in such a way, that users must use the satellite ecosystem (whish is not a part of Usenet) exclusively to download them, when the website disappears - the encrypted scattered posts just eat usenet providers' disk space uselessly. If you can't find something specific please let me know what exactly you can't find for me to check how it is possible to retain the posts. Legible posts shouldn't be affected at all, let me know if you notice anything missing.
08 Aug 24: After being unchanged for many years the search service communication protocol had to be updated a few times in part to resolve a server side set feature limitation and properly handle trial searches on the relocated server. The searches themselves are not affected and it looks like the work is now complete.
November 13: A server which was supposed to be upgraded was attacked by ransomware, probably due to rdp vulnerability, the search service is currently down for backup.
November 14: I'm restarting the search service with limited retention. The affected server had been already scheduled for upgrade, but I will need now also to copy the backup data to it which will take some time.
November 20: The server affected by the ransomware attack had the system reinstalled and it has been upgraded (RAM, disk space). I'm in the process of copying data from another server. Even as there is more RAM, for a subset of groups I will try to run mid retention range in the compact mode using sets as the server then starts several times faster and consumes twice less RAM, we'll see how it goes.
December 1: I will be engaging the server soon, maybe will try later today or tomorrow as it may take time to convert the database into the compact mode. It may take another week to copy all the data, but maybe I will be able already to run it very near full retention including the oldest headers.
December 2: The restored server has been engaged. There is a retention gap for some groups between 1200-2300 days, but it will be filled within a few days. Some server instances are run in the compact mode (natively supported sets) which wasn't a case in the past. Overall retention range is around 5300 days.