On Sat, 2009-09-19 at 02:23 -0400, Aaron Wolfe wrote: > 2009/9/18 Karsten Bräckelmann: > > This machine NEEDS more RAM. In fact, I'd guess half of the spam > > slipping through is due to timeouts. Thrashing into hell. > > throwing ram at a server is not a solution in this case. 512MB is > sufficient to handle this mail load, as indicated by his post showing > little swap utilization on the system and confirmed by my real world
You're right, Aaron, the output of 'free' suggests this is not actually a problem. Alas, even though I asked repeatedly, this data point was given after that post of mine, and I was limited to very little info and some observations. > experience. here we handle over 1 million messages per day per node, > each node has 1GB ram. ram required is easily calculated by base > services + SA instance usage X number of instances you'd like to use. > having less instances generally just means slight (very slight in most > cases) delays. having more instances than your ram can contain means > big delays. properly configured server will not start swapping and > falling over when a flood of mail comes in, mail simply spends more > time in queue. the difference between 1 second and 1 minute in queue > is not usually significant to users. > > the problem here is bad administration. hopefully with the advice > given on list and better yet some time spent studying docs, this can > be corrected. -- char *t="\10pse\0r\0dtu...@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4"; main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1: (c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}