[nycbug-talk] Stablity fixed??? Maybe???

[George R.]

Matt Juszczak wrote:
> Hi all,
> 
> Part of this message I cross posted to freebsd-questions, but no one 
> there is going to give me a good answer ... so I'd like a 2nd opinion :)
> 
> After removing IPF and having a week of stability, we decided to put
> our mail server to the test today.
> 
> I began flooding it with tons of mail messages using smtp-source at 
> about 2 pm today.  The server load jumped up to about 4.50 average.  It 
> eventually started denying requests, but after waiting a few minutes, it 
> would accept them again (I was literally flooding it, I sent in all 1.2 
> million emails).
> 
> I wrote a scipt to hammer it, so even after it would refuse the connection,
> it would hammer it again.  Eventually, the machine started to not respond.
> I could ping it with successful replies, but could not SSH into it.  The
> last message on the screen was "Could not write to /var/mail/thissucks",
> which was the account we were testing.
> 
> I rebooted the machine, and all is fine.  I'm not sure if this is still a
> sign of instability, or if this is a "Any idiot who would sent 1.2 million
> emails and a full flood for hours to a mail server should expect something
> like this to happen" message.

I assume you're tail'g /var/log/maillog or messages while this is 
happening. . . might be useful to know what's listed there. . .

Did you try to telnet to 110. . . this would be useful since you'd 
obviously be seeing what the hypothetical remote box would see.  I'd 
assume it would entertain telnet before ssh requests. . .

I followed your thread (s) and just wanted to say, i had some bad 
experiences recently also with FBSD <=5.3, but one time is was (brand 
new) bad RAM. . .  I would do the make install world ladi-doti, then 
make would bomb out at different times, but the OS seemed fine overall. 
  Memtest showed me the light though. . . which was probably the only 
reasonable explanation to arbitrary crashes during make. . .

g