Closed Queeq closed 8 years ago
We have added compression and (optional) encryption some time ago. Can you please confirm the versions of ntopng and nprobe you are using?
nprobe 7.3.160419-5042 ntopng 2.3.160419-1164
Please send a pcap file (full packet size) with flows sent by your routers to nProbe as I need to debug the issue. Note that I need both flows and templates.
We use default ipv4-template by Juniper. It is described here. I'll send sample flows over e-mail.
I tested it yesterday and it worked. I think we may close this issue now. I wonder if latest commits resolved #507 too.
Seems like I was too fast closing it. It had been working for some time but stopped working after the first restart. I'm running nprobe 7.3.160422-5045 and ntopng 2.3.160422-1178. Symptoms are the same.
Without a pcap I can't help much
@lucaderi Please check your e-mail, I sent it to the e-mail in your Github profile on the 19th of April, 15:10 UTC. The subject was "ntopng issue #505 flows pcap".
@Queeq Look like the pcap you sent me contains flows but not templates. Please try again.
Closing for inactivity. Will reopen if necessary.
@lucaderi Sorry, I was on vacation during the last days. I've just (minutes ago) managed to include templates into the Netflow packets from our routers and will shortly send the traffic dump by e-mail to you.
For now, just for reference for other people exporting IPFIX from Juniper routers, the following is the configuration that is necessary to include templates together with flows:
show configuration services
flow-monitoring {
version-ipfix {
template NAME {
flow-active-timeout 30;
flow-inactive-timeout 60;
template-refresh-rate {
packets 1;
seconds 10;
}
ipv4-template;
}
}
}
Note that you need both packets
and seconds
options enabled within template-refresh-rate
for it to work.
@lucaderi Have you received the pcap file on April 28th? Please reopen if yes.
Running latest nightlies on VMWare virtual machine Ubuntu 14.04, kernel 4.4.6.
There are 6 nprobe instances receiving IPFIX from Juniper routers and 1 ntopng instance which connects to nprobes over TCP (everything's within the same VM). It had been working fine on the same version before I restarted it several times today (both nprobes and ntopng).
Nprobes are run with these parameters:
Ntopng verbose mode shows many messages like these:
Sniffing traffic on loopback interface gives this:
Note packet length of 8192 from nprobe. During normal operation I noticed it was sending packets 150-200 bytes long.