4 ms·
Hackernews hug of death already?
by the_duke 7y ago
Hackernews hug of death already?
- blaisio 7y agoMaybe they should use this module that I can't read about to power their site. :P
- bdcravens 7y agoI assume the downtime is due to the HTTP server, not persistence.
- falcolas 7y agoThat would be unusual. Most websites fall over due to a DB bottleneck long before the backends or HTTP servers see significant load.
- paulddraper 7y agoYep. Almost always the web server will not be able to connect to the db because of the limited connections + slowed db. --- FWIW, the site is up for me.
- jdsully 7y agoThe bulk of the CPU time is currently spent in node.js, running docusaurus.
- mpweiher 7y agoMakes me feel a little better about my libµhttp/Objective-C/Objective-Smalltalk server (serving http://objective.st http://objective.st), which held up just fine to the HN hug of death, running on the smallest digital ocean droplet. :-)
- quickthrower2 7y agoThose $5 droplets pack a surprising punch.
- majewsky 7y agoA simple nginx also holds up very well: https://xyrillian.de/thoughts/posts/latency-matters-aftermath.html https://xyrillian.de/thoughts/posts/latency-matters-aftermat...
- mamon 7y agoSeems that even that doubled Redis performance wasn't enough.
- jdsully 7y agoThe AWS instance was sized too low :) touche on the not fast enough comments.
- brodouevencode 7y agoTesting to see if we get permabanned for using the term "slash-dotted"