wardb
Joined 15 years ago
Joined 15 years ago
The linked Loki caching docs/articles are for optimising the read access patterns of S3/object storage, not for writes.
The difference is that in Loki the index is only used for metadata around the source of the log lines (environment, team, cluster, host, pod etc) for selecting the right log stream to search in.
Parsing, aggregation and/or filtering of log lines on high cardinality data is all done at query time using LogQL. See also https://www.youtube.com/watch?v=UiiZ463lcVA and this live example where a 95th quantile is calculated using the request_time field of nginx logs https://play.grafana.org/d/T512JVH7z/loki-nginx-service-mesh...
The expense categorization system is an abomination that will drive you insane with its constant mistakes and errors. The approval process is a black hole that sucks up your expenses and never spits them back out. I feel SAP Concur is the worst software ever created for expense management, and using it makes me wish I never had to do expenses again.
Expensify is so much better. The OCR isn't great though
High availability in Loki is supported in distributed mode. Helm chart here: https://github.com/grafana/helm-charts/tree/main/charts/loki...