Hi, I think I know this answer, but would like to be sure.

Is there any way to force a reload of a pattern file without a new
haproxy process needing to be spawned? Since stick tables can't be
used for storing XFF IP addresses, I've concocted my own scraper
detection external to haproxy and I'm delivering the IP addresses to a
haproxy via a pattern file.

Right now I'm reloading haproxy ~2 mins which loses backend health
state and stats. I'd like to reload as often as new IP's are
discovered by my 'Rube Goldberg' scraper detection, but that would be
as fast as 30 seconds which is too often given that backend health
detection takes a significant fraction of that time.

Reply via email to