reset the rb to factory (glutton for punishment) and it's no longer crashing. yay. my mikrotik devices have been super stable i've never had luck with any of the ones i've used i've had this 450 for like 5 years, and stopped using it because each bugfix release broke new stuff and mikrotik generally doesn't care also, i was one of the lucky people who got a device out of the batch w/bad capacitors in it (they ignored reports of devices blowing up for almost 6 months before admitting they had a mfg issue, but refused to warranty repair anything) i have a RB751G-2HnD and a CRS125-24G-1S both newer devices than yours i have an rb450g and used to have an rb1100ahx2 the 450 is one of their most-sold devices for WISPs and stuff given the option, i'd still use one over any consumer stuff agreed and their qos flexibility is great i'm not doing anything complex on mine so it probably explains the stability all i did was qos and dumb nat stuff vpn functionality on the 1100 was godawful, so i just gave up on trying to do that and sold it VPN stuff is much better now in the 6.x series memory use is better on 6 too i've only had the RB751G-2HnD crash once and reboot never seen it again this one was doing all kinds of squirrelly shit last night before i reset it to factory, haha it would crash if i disabled qos rules, it would crash again when enabling them, it crashed when i enabled dhcpv6 client, crashed again when i configured it to RA the ipv6 prefix it got from comcast just weird all around guessing the upgrade from 5.x to 6.x didn't like an older config or something on the device since wiping stabilized it yeah there are bugs upgrading from 5.x to 6.x they fixed a few of them in recent 6.x releases they look intersting from hardware point of view it's the software side that makes me apprehensive tilera cpus sound like they're pretty good but i don't htink routeros is taking proper advantage of it yeah it isn't i don't think the x86 stuff is quite there yet for routeros x86? yeah the tilera stuff is x86 oh i thought it was more arm like actually on their page it says Architecture TILE http://routerboard.com/CCR1036-8G-2Splus there are cards for pc with tile cpus on them maybe that made you think x86? yeah but iirc it still does BGP on one core see something like that looks good until you realise it can't do simple mirror functionality mirror port bgp on one core shouldn't be that bad juniper still only use one core for bgp too :/ now im trying to figure how why php-fpm is giving me a blank page i had that happen to me too i can't quite remember what it was at that point though and its hard to debug cause i can't figure out how to make php-fpm return an actual error yeh that's what i thought check the nginx error log? and fpm error log? maybe it'll come back to me what happened to me haha did you know php-fpm can backtrace a page that takes too long to load? nothing in the fpm error log, nginx error log, php-fpm is returning 500s 2001:470:1d:96b::1000 - - [06/Dec/2014:18:51:52 +0000] "GET / HTTP/1.1" 500 5 "-" "Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/39.0.2171.71 Safari/537.36" i think what i stumbled on for longest was that i was using alias rather than setting root and files were trying to get served from wrong place. that might be it but I can't see it i had something in nginx error log in the end i symlinked cacti to /usr/share/cacti/site cos it was cacti using php it kept trying to load /usr/share/cacti/site/cacti/ when going to hostname/cacti with lighttpd i just used aliases. https://gist.github.com/staticsafe/294c4b56ef5b57de2048 Gist: "asininetech.com.conf" oh i think you have to stick the .php thing inside the location? or at least before the location oh it says it matches last all? err matches last when screwing around i stuck my .php location thing before the normal location thing and now i dunno if it's necessary or not yeah its the config I copied from the currently working server that I'm planning on moving ahh my other host doesn't. and you checked socket permissions? its using a tcp socket hmm i have fastcgi_pass unix:/var/run/php5-fpm.sock; oh upstream php { server 127.0.0.1:9000; } yeah and that's listening fine? tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN 29594/php-fpm: mast my php fpm has like nothing useful at all generally [23-Nov-2014 10:36:05] WARNING: [pool www] child 15800 exited on signal 11 (SIGSEGV - core dumped) after 87.360830 seconds from start don't remember what that was but it was ages ago current server is Debian, this new one is CentOS 7, i wonder if there is some small difference breaking things i had problems with ubuntu but not arch :) try tcpdump tcpdump -l -n -i lo port 9000 oh good idea fwiw, php-fpm dumps errors to the server's error log. But if you're getting 500, that's a server configuration issue, not php-fpm's. hm (not that it's impossible for php-fpm to return a 500 too, as it would in this case. But the fact that there's nothing in the server error log suggests that it's not php-fpm's fault. If it were php-fpm, I'd expect something like: 2014/12/06 11:40:14 [error] 4925#0: *1332261 FastCGI sent in stderr: "PHP message: PHP Parse error: syntax error) static any luck? mercutio: nope i'm giving up for now ahh ok Just finished walking my dad through restoring his malware infected Dell Windows 7 machine back to factory settings (over the phone) i see news.arpnetworks.com is slowly getting a few more users.