Post B6GcunbwgwvG4jZTqy by mdione@en.osm.town
(DIR) More posts by mdione@en.osm.town
(DIR) Post #B6GcunbwgwvG4jZTqy by mdione@en.osm.town
0 likes, 0 repeats
I figured out (with some help) how to use an SQL database to monitor #apache with #Grafana, skippimg Prometheus or Loki, which (I think) would explode trying to handle IPv4s, URLs, refrers or User Agents:https://www.grulic.org.ar/~mdione/glob/posts/monitoring-apache-with-sql-and-grafana/#monitoring #SRE #HomeLab
(DIR) Post #B6Gcunp3uBPkjPhxIW by feld@friedcheese.us
0 likes, 0 repeats
@mdione > skippimg Prometheus or Loki, which (I think) would explode trying to handle IPv4s, URLs, refrers or User Agents:because of high cardinality? It would probably be fine with VictoriaMetrics, that's a prometheus implementation that handles high cardinality very efficiently
(DIR) Post #B6Gcuo2X66BpPC0iIK by mdione@en.osm.town
0 likes, 0 repeats
@feld oh, wow, good to know. But would the be easy to answer questions like:* count of status codes, user agents, etc per minute.* total data transferred (rx, tx) per minute.* all the logs in a given time interval.* min/max response time.* count of request per client (per IP).* list of URLs per client (assuming a variable has the client IP).* count of requests per client* list of URLs with a 4xx status.This is what I have in my dashboard so far, SQL is so flexible!
(DIR) Post #B6GcuoLJyFDcLSnia0 by feld@friedcheese.us
0 likes, 0 repeats
@mdione SQL might be more flexible, don't rearchitect if what you have is working. Logs isn't its strong suit, but they have a logs product too...Just keep Victoria in your mind if you have a new project that needs Prometheus compatible TSDB which is very performant and considerably lighter than the official implementation.(it also supports: OpenTSDB, Influx, and more)
(DIR) Post #B6GcuoVxKhj2sRmD9k by hj@shigusegubu.club
0 likes, 0 repeats
@feld @mdione heheperse