[c-nsp] Error Msgs

[Paul Stewart]

Thanks for the feedback.... it's happening every hour or two in the logs....
 
I have adjusted the renewal time on the DHCP though - we had set it to one
hour while getting things online - changed it to 24 hours hoping that this
will alleviate the issue possibly....  it's quite possible that all 1000+
devices are renewing their IP's at exactly the same time currently - this
will change over time but they were all brought online at the same time....
 
If there's anything I can adjust to work around this I'd love to hear back -
these devices were being served via a 6500 at one point and didn't have any
issues ...
 
Thanks,
 
Paul
 

  _____  

From: Karol Mares [mailto:karol.mares at gmail.com] 
Sent: Sunday, September 02, 2007 9:41 AM
To: Paul Stewart
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Re: Error Msgs


Hi,


On 9/2/07, Paul Stewart <paul at paulstewart.org> wrote: 

Hi there....

There's lots of memory free on the box...

There's about 1000 devices connected to it - do I need to fine tune buffers?
Also, am providing DHCP services to all these devices...

Thanks!

dis2-rtr-tu#sh processes memory sorted
Processor Pool Total:  948488604 Used:   20778360 Free:  927710244
      I/O Pool Total:   39845376 Used:   14150160 Free:   25695216

PID TTY  Allocated      Freed    Holding    Getbufs    Retbufs Process 
   0   0   57090732   16619932   28420708          0          0 *Init*
130   0    1570744        428     974472          0          0 DHCPD
Receive



>From what i can see here , it is a IP Input process.Chunk can not be
allocated when the box is running out of memory, the memory is fragmented
from the malloc(),  free() functions or if the malloc() function wants to
allocate 0 size of chunk. It is also possible that the memory was bit
fragmented and there were many DHCP requests send on the box. 

I have seen silimiar chunk problem on 3825 running EzVPN server

--
         iso