Hi Nate
please note that a dummy interface is a kernel interface with limited 
performance. This regardless of the consumers actually running (when
you run a consumer, you just add load increasing the packet loss), as
zbalance_ipc is sending traffic to the kernel anyway.

Best Regards
Alfredo

> On 28 Jan 2019, at 17:46, Alexander, Nathan 
> <[email protected]> wrote:
> 
> 
> I am having an issue with zbalance_ipc where we are dropping 10% of packets 
> at 1.5Gbps+/interface  even before any consuming applications begin to 
> consume the data.  The setup is a 2x18 core server where one application is 
> using a single queue and the other application is consuming 18 queues, and we 
> have a zbalance process per NUMA Node.  When we start z_balance with only one 
> application (-n 1), we see no loss but with -n 18,1, we see ~10% packet drops 
> before anything is hooked to the dummies:
>  
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:266] Absolute Stats: Recv 1'547'957'308 
> pkts (138'616'411 drops) - Forwarded 3'095'914'616 pkts (0 drops)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:305]                 zc:p1p2 RX 
> 1547957313 pkts Dropped 138616411 pkts (8.2 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 0 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 1 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 2 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 3 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 4 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 5 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 6 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 7 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 8 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 9 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 10 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 11 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 12 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 13 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 14 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 15 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 16 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 17 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:319]                 Q 18 RX 0 pkts 
> Dropped 0 pkts (0.0 %)
> 23/Jan/2019 21:58:59 [zbalance_ipc.c:338] Actual Stats: Recv 256'628.26 pps 
> (0.00 drops) - Forwarded 513'257.53 pps (0.00 drops)
>  
> The interfaces are intel NICs, 2x10GE but we're only consuming data of of one 
> interface on each card.
>  
> 0% packet loss (running as background so that I can get logs, but normally 
> run with -d):
> sudo /usr/bin/zbalance_ipc -i zc:p1p2 -a -q 131072 -c 99 -n 1 -m 1 -S 4 -g 2 
> -r 0:dummy57 -l /var/log/zbalance_p1p2.log -v -p -u /dev/hugepages &
> sudo /usr/bin/zbalance_ipc -i zc:p4p2 -a -q 131072 -c 97 -n 1 -m 1 -S 5 -g 3 
> -r 0:dummy59 -l /var/log/zbalance_p4p2.log -v -p -u /dev/hugepages &
>  
> 10% Packet drops to start, greater with actual consumers on the dummies:
>  
> sudo /usr/bin/zbalance_ipc -i zc:p1p2 -a -q 131072 -c 98 -n 18,1 -m 4 -S 8 -g 
> 6 -r 0:dummy20 -r 1:dummy21 -r 2:dummy22 -r 3:dummy23 -r 4:dummy24 -r 
> 5:dummy25 -r 6:dummy26 -r 7:dummy27 -r 8:dummy28 -r 9:dummy30 -r 10:dummy31 
> -r 11:dummy32 -r 12:dummy33 -r 13:dummy34 -r 14:dummy35 -r 15:dummy36 -r 
> 16:dummy37 -r 17:dummy38 -r 18:dummy57 -l /var/log/zbalance_p1p2.log -v -p -u 
> /dev/hugepages &
>  
> sudo /usr/bin/zbalance_ipc -i zc:p4p2 -a -q 131072 -c 96 -n 18,1 -m 4 -S 9 -g 
> 7 -r 0:dummy0 -r 1:dummy1 -r 2:dummy2 -r 3:dummy3 -r 4:dummy4 -r 5:dummy5 -r 
> 6:dummy6 -r 7:dummy7 -r 8:dummy8 -r 9:dummy10 -r 10:dummy11 -r 11:dummy12 -r 
> 12:dummy13 -r 13:dummy14 -r 14:dummy15 -r 15:dummy16 -r 16:dummy17 -r 
> 17:dummy18 -r 18:dummy59 -l /var/log/zbalance_p4p2.log -v -p -u 
> /dev/hugepages &
>  
> p4p2 is hooked to NUMA 1
> P1p2 is hooked to NUMA 0
> Other relevant data from the server:
> NUMA node0 CPU(s):     
> 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70
> NUMA node1 CPU(s):     
> 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71
>  
> ixgbe.conf:
> RSS=1,1,1,1,1,1 numa_cpu_affinity=2,2,2,2,3,3
>  
> PF Ring version is:
> zbalance_ipc -h
> zbalance_ipc - (C) 2014-2018 ntop.org <http://ntop.org/>
> Using PFRING_ZC v.7.3.0.180817
>  
> Let me know if I'm missing something obvious or any ideas on how to improve 
> performance.  Thanks so much!
> 
> Cheers,
>  
> Nate
> 
> 
> 
>  
>  
> 
> NOTICE: Morgan Stanley is not acting as a municipal advisor and the opinions 
> or views contained herein are not intended to be, and do not constitute, 
> advice within the meaning of Section 975 of the Dodd-Frank Wall Street Reform 
> and Consumer Protection Act. If you have received this communication in 
> error, please destroy all electronic and paper copies and notify the sender 
> immediately. Mistransmission is not intended to waive confidentiality or 
> privilege. Morgan Stanley reserves the right, to the extent required and/or 
> permitted under applicable law, to monitor electronic communications, 
> including telephone calls with Morgan Stanley personnel. This message is 
> subject to the Morgan Stanley General Disclaimers available at the following 
> link: http://www.morganstanley.com/disclaimers 
> <http://www.morganstanley.com/disclaimers>.  If you cannot access the links, 
> please notify us by reply message and we will send the contents to you. By 
> communicating with Morgan Stanley you acknowledge that you have read, 
> understand and consent, (where applicable), to the foregoing and the Morgan 
> Stanley General Disclaimers.
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Ntop mailing list
> [email protected] <mailto:[email protected]>
> http://listgateway.unipi.it/mailman/listinfo/ntop 
> <http://listgateway.unipi.it/mailman/listinfo/ntop>
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to