I am seeing many such errors in the error log;
failed (104: Connection reset by peer) while reading response header from upstream, client: 192.168.0.168, server: _, request: “GET /index.php?sp_ahah=subs-manage&target=subs&sfnonce=75a0081e56&rnd=1349339737820 HTTP/1.1”, upstream: “fastcgi://127.0.0.1:9005”,
Anyone has had similar problem and managed to know the cause?
no, afraid not… I would talk to my host and see if they can provide any insight… its not a critical action… just the update of the current unread subscription count…
do you get lots of those? or just a stray one… you could turn off the auto update on forum – options – global settings to prevent them… just would mean that the subs count, watches count and inbox count would only get updated on page loads…
Thanks! I am my own host.. yeah i get those errors when the forum starts to get traffic?.. i think i can live without autoupdate for now and play with the bug when i have more time. There are tons to watch out when Nginx is running … such as 502 gateway errors that i am trying to understand.
I did that… up the mem and set higher timeout. Not a big problem.. Now it only happens when i abusively open 10+ windows of the same site at the same time. not a bad thing instead – 502 gateway errors to ddos attacks. 
cron did run for some reason after i restored the site from a fresh backup.. i am wondering if it’s normal to see the Next Run (date) before the current time/date. Manually run the hook doesn’t update the next run date nor wp-cron.php.
no,not normal… but if wp cron wasnt working, then the dates could have gotten screwed up… would think wp would sort that itself…
so go ahead and deactivate sp plugin… then reactivate… should at least reset our crons…