Tootfinder

After months without working search, I deleted the elasticsearch 8 container and created a new one with elasticsearch 9. Currently hydrating it with tootctl search deploy. 12.721.688 documents, 2 hours left.
I hate self hosting. (Still thankful to everyone who makes it possible.)

test toot for elasticsearch

Elastic Meetup #98 - Une command line pour #Elasticsearch : Oxydation d'un process
avec l'ami @…
https://w…

Today I spend a few hours fixing a server crash in my project with the most users. We moved the app to a new server yesterday. Today at lunch it crashed (right during Mittagsjournal..). The log was full of ElasticSearch errors complaining about 6278198379712 being bigger than maxint. Fair enough. After some digging I found searches requesting this hugh number as the page param. Which makes no sense, we don't have that much content. The quick fix was to limit page to 100. Then the proper …

⚡ Uses up to 30x less RAM and 15x less disk space than Elasticsearch and Grafana Loki. Capacity and performance scale linearly with available resources - runs smoothly on Raspberry Pi and on servers with hundreds of CPU cores and terabytes of RAM. Scales horizontally to hundreds of nodes in cluster mode.

Today we had the first day of the Koha DACH Hackfest in Göttingen. After a tour through the data center we started various discussions on Koha. The first feature that was requested was something I actually implemented some time ago (passing SQL to the ElasticSearch indexer), but failed to add to a weird wrapper script I did not know existed. Which I primptly fixed:

Getting Your Indices in a Row: Full-Text Search for LLM Training Data for Real World
Ines Altemir Marinas, Anastasiia Kucherenko, Alexander Sternfeld, Andrei Kucharavy
https://arxiv.org/abs/2510.09471 …

Getting Your Indices in a Row: Full-Text Search for LLM Training Data for Real World
The performance of Large Language Models (LLMs) is determined by their training data. Despite the proliferation of open-weight LLMs, access to LLM training data has remained limited. Even for fully open LLMs, the scale of the data makes it all but inscrutable to the general scientific community, despite potentially containing critical data scraped from the internet. In this paper, we present the full-text indexing pipeline for the Apertus LLM training data. Leveraging Elasticsearch parallel i…

🔧 Zero-config single executable setup - no tuning required for VictoriaLogs or the operating system. Reasonable defaults automatically adjust for available CPU and RAM resources. Much easier to set up and operate compared to Elasticsearch and Grafana Loki.
🔌 Accepts logs from popular collectors including Fluentbit, Fluentd, Logstash, Vector, Promtail, #OpenTelemetry, DataDog Agent…