this post was submitted on 21 Jul 2023
203 points (98.1% liked)
Fediverse
28465 readers
650 users here now
A community to talk about the Fediverse and all it's related services using ActivityPub (Mastodon, Lemmy, KBin, etc).
If you wanted to get help with moderating your own community then head over to !moderators@lemmy.world!
Rules
- Posts must be on topic.
- Be respectful of others.
- Cite the sources used for graphs and other statistics.
- Follow the general Lemmy.world rules.
Learn more at these websites: Join The Fediverse Wiki, Fediverse.info, Wikipedia Page, The Federation Info (Stats), FediDB (Stats), Sub Rehab (Reddit Migration), Search Lemmy
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Every one minute seems... excessive. I know this is social media but still, what would be bad about reducing that to like once per hour, to reduce the load across the entire Fediverse?
I think you're underestimating the number of requests that a server can handle. Even my tiny instance currently sees dozens of requests every second and is very lightly loaded. A single request per minute is an immeasurably small load.
You made me curious. Since midnight lemmy.ca has averaged 110 requests per second, with multiple periods of doing ~3000 request/sec.
1 request every minute is nothing :)
Every one minute is really nothing
A single health check call a minute by a single app is nothing compared to real world usage
True, but if we agree that every second is too frequent, but that every day is too sparse (maybe?), then the question becomes one of optimizing the choice of what frequency would be best for whatever outcome.
Like if it was every 5 minutes, that would cut down on the amount of traffic 5-fold, and still give 12 checks every hour. If we do not expect checks to fail, that may still be 6-12-fold more often than necessary. Or a check every 10 minutes is still "many checks per hour", and yet does not blow up the traffic stats 10-fold more than necessary to get a "desirable" outcome.
There are also additional costs to consider when doing a check every minute: that is 1440 checks every 24 hours! Data must be stored, CPU time spent, waste heat is generated (it is not exactly exasperating climate change but it is not not doing that either!), etc. Whereas if a check every 10 minutes is used, then the same number of checks would span a 10-day timeframe. Depending on whatever else the machine is doing (gaming? other server checks?), it reduces the load 10-fold and something that much may even change things on a qualitative rather than merely quantitative basis.
Well, it was a thought for consideration at any rate.