It doesn't take much unbalancing ---
the hash grows 32 -> 48 -> 72 ->
if you have 32 on one side and 48 on the other and conditions are just
"right"... boom!

-----Burton

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf
Of Luis Vazquez
Sent: Wednesday, January 16, 2002 8:31 AM
To: '[EMAIL PROTECTED]'
Subject: RE: [Ntop-dev] To all those having problems of Ntop just
aborting (2.0)


for me it happens either afteer 30 minutes or after 1 day. I have on that
box 1 GB of RAM + 512 Mb of swap disk space, I don't think the memory would
be the problem, my network is not unbalanced however I'm using the -M switch
that I just removed. I'll let you know if it happens again.

Regards

Luis


-----Original Message-----
From: Burton M. Strauss III [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, January 16, 2002 9:14 AM
To: Ntop; Ntop-Dev
Subject: [Ntop-dev] To all those having problems of Ntop just aborting
(2.0)


That is, those of you who are having NTop shutdown without explanation and
you're sure you have enough memory.

Some of you are seeing Ntop fail and swear you aren't memory constrained
(I'd really ask you to try shutting down un-necessary daemons, at least to
show it isn't memory).  The failure I'm looking at is pretty predictable -
in fact, for Chris Picton it happens after about 60 seconds, 100% of the
time...

We have demonstrated (thanks Rafel) a different problem in a severely memory
constrained environment (this one goes away if you shut down other daemons
running on the box, which allows NTop to get "enough" memory).

With a lot of help from Chris Picton running test patches for me, I
***THINK*** I see the problem, I just need to confirm it.

If you fit the picture... I would like to confirm two things;

1. Whether you are using the -M flag
2. Whether your network is "unbalanced" - that is two (or more) interfaces
with a different number of hosts on either side. If so, in the NTop log, you
will see Hash (not TCP hash) Extending messages ("15/Jan/2002 17:01:01
Extending hash: [old=32, new=48]") and also Index error messages, e.g.
"15/Jan/2002 17:01:04 Index error idx=37 @ [pbuf.c:493]"
3. If you are using -M, does the problem go away if you stop using -M?

If you can grep those messages out of the logs (DON'T SEND THE WHOLE THING),
I'd appreciate it!

Thanks!

-----Burton

_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listmanager.unipi.it/mailman/listinfo/ntop-dev
_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listmanager.unipi.it/mailman/listinfo/ntop-dev

_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listmanager.unipi.it/mailman/listinfo/ntop-dev

Reply via email to