[Info-vax] TCPIP tying up system

Steven Schweda sms.antinode at gmail.com
Sun Nov 28 12:05:39 EST 2010


Phillip Helbig---undress to reply wrote:
> Recently, I've noticed a slowdown [...]
> [...]
> OpenVMS V7.3-2  [...]

   So that's an Alpha?

>   Pid    Process Name    State  Pri      I/O       CPU       Page flts  Pages
> 24060604 TCPIP$SMT_BG416 COM     11      152   0 00:00:18.90      3233    240  N
> 24060E06 TCPIP$SMT_BG452 COM     11      174   0 00:00:21.98   1003864    281  N
> 24061015 TCPIP$SMT_BG461 COM     11      104   0 00:00:20.67   1002070    187  N

   Lots of "Page flts", not many "Pages".  Around here, for
example:

  Pid    Process Name    State  Pri      I/O       CPU       Page
flts  Pages
202022D2 TCPIP$S_BG59299 LEF      4      577   0 00:00:05.74
18809   8942  N

> MONITOR showed SWAPPER taking the largest CPU share, about 37%.
>
> I suspect some sort of spam flood or DOS attack.

   I wouldn't call ten of anything much of a flood.  I suspect
not enough memory (or not enough memory for those who need
it).  Assuming that you have some free physical memory, you
might try letting Username: TCPIP$SMTP see a little more of
it.  Around here, for example:

      WSquo:         8192
      WSextent:     16384
      Pgflquo:     204800

I don't remember what the initial values were, and I don't
claim that these are optimal, but my "Page flts" seem to be
more reasonable than yours.  What did you do, overlay a VAX
installation (with its tiny memory requirements) with an Alpha
OS (with its significantly larger memory requirements)?  Or
are the default values really that small?  Or are you trying
to get by with 64MB of physical memory in this system?

   Have you run AUTOGEN lately?



More information about the Info-vax mailing list