Closed GoogleCodeExporter closed 9 years ago
Are you still getting high UDP packet losses? I haven't seen this. Possibly
it got fixed along with the recent performance improvements.
Original comment by jef.posk...@gmail.com
on 1 Dec 2013 at 2:15
Yes, I still see 88% packet loss.
iperf3 -u -b1G -c 192.168.102.9
Connecting to host 192.168.102.9, port 5201
[ 4] local 192.168.102.8 port 35631 connected to 192.168.102.9 port 5201
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-1.00 sec 116 MBytes 971 Mbits/sec
[ 4] 1.00-2.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 2.00-3.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 3.00-4.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 4.00-5.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 5.00-6.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 6.00-7.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 7.00-8.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 8.00-9.00 sec 128 MBytes 1.07 Gbits/sec
[ 4] 9.00-10.00 sec 128 MBytes 1.07 Gbits/sec
Original comment by bltier...@es.net
on 1 Dec 2013 at 3:51
Take.
I've recreated this problem with 40G hosts on the testbed with the tip of the
mainline, no significant changes from Comment 2:
[bmah@nersc-diskpt-6 ~]$ ./iperf3 -u -b1G -c 192.168.101.9
Connecting to host 192.168.101.9, port 5201
[ 4] local 192.168.101.8 port 56236 connected to 192.168.101.9 port 5201
[ ID] Interval Transfer Bandwidth Total Datagrams
[ 4] 0.00-1.00 sec 116 MBytes 972 Mbits/sec 14830
[ 4] 1.00-2.00 sec 128 MBytes 1.07 Gbits/sec 16384
[ 4] 2.00-3.00 sec 128 MBytes 1.07 Gbits/sec 16384
[ 4] 3.00-4.00 sec 128 MBytes 1.07 Gbits/sec 16384
[ 4] 4.00-5.00 sec 128 MBytes 1.07 Gbits/sec 16384
[ 4] 5.00-6.00 sec 128 MBytes 1.07 Gbits/sec 16381
[ 4] 6.00-7.00 sec 128 MBytes 1.07 Gbits/sec 16384
[ 4] 7.00-8.00 sec 128 MBytes 1.07 Gbits/sec 16385
[ 4] 8.00-9.00 sec 128 MBytes 1.07 Gbits/sec 16383
[ 4] 9.00-10.00 sec 128 MBytes 1.07 Gbits/sec 16384
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Jitter Lost/Total
Datagrams
[ 4] 0.00-10.00 sec 1.24 GBytes 1.06 Gbits/sec 0.001 ms 138083/162214
(85%)
[ 4] Sent 162214 datagrams
iperf Done.
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.101.8, port 60520
[ 5] local 192.168.101.9 port 5201 connected to 192.168.101.8 port 56236
[ ID] Interval Transfer Bandwidth Jitter Lost/Total
Datagrams
[ 5] 0.00-1.00 sec 16.1 MBytes 135 Mbits/sec 0.002 ms 12743/14805
(86%)
[ 5] 1.00-2.00 sec 19.7 MBytes 165 Mbits/sec 0.002 ms 13861/16381
(85%)
[ 5] 2.00-3.00 sec 19.7 MBytes 165 Mbits/sec 0.001 ms 13876/16396
(85%)
[ 5] 3.00-4.00 sec 19.7 MBytes 165 Mbits/sec 0.002 ms 13859/16379
(85%)
[ 5] 4.00-5.00 sec 19.7 MBytes 165 Mbits/sec 0.001 ms 13866/16386
(85%)
[ 5] 5.00-6.00 sec 18.6 MBytes 156 Mbits/sec 0.009 ms 14018/16400
(85%)
[ 5] 6.00-7.00 sec 19.0 MBytes 160 Mbits/sec 0.001 ms 13864/16301
(85%)
[ 5] 7.00-8.00 sec 18.7 MBytes 157 Mbits/sec 0.001 ms 14001/16394
(85%)
[ 5] 8.00-9.00 sec 18.6 MBytes 156 Mbits/sec 0.001 ms 13998/16384
(85%)
[ 5] 9.00-10.00 sec 18.7 MBytes 157 Mbits/sec 0.001 ms 13997/16388
(85%)
[ 5] 10.00-10.00 sec 0.00 Bytes 0.00 bits/sec 0.001 ms 0/0 (-nan%)
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Jitter Lost/Total
Datagrams
[ 5] 0.00-10.00 sec 1.24 GBytes 1.06 Gbits/sec 0.001 ms 138083/162214
(85%)
Original comment by bmah@es.net
on 8 Jan 2014 at 11:17
Close as invalid for now. After considerable testing, we note that we've only
observed this problem on Mellanox interfaces in the ESnet 100G testbed, and
there might be some correlations with other issues observed with other
experiments. However, running on the same hardware, we have not observed any
abnormal loss on Myricom or Intel (e1000e) type interfaces.
Original comment by bmah@es.net
on 17 Jan 2014 at 8:41
Original issue reported on code.google.com by
bltier...@es.net
on 26 Nov 2013 at 4:01