Hi Ramesh, Despite using a little unlucky term for it, this is a completely normal event in a link. It just means that the link send queue is temporarily full, and returns -EAGAIN to the calling socket. The socket will then sleep until it receives a 'congestion abated' callback from the link, or it will return -EAGAIN to the calling socket user if the socket is non-blocking.
You won't see any impact on the user application, because this statistically will be an infrequent event occurring only during high traffic. Neither is duplicate packets anything to worry about, -those are detected and dropped without affecting the overall packet flow. The indication you see is just statistics, helping the reader to keep track of the number of redundant re-transmissions, and hence assess the quality of the carrying media. BR Jon Maloy On 10/31/2016 02:00 PM, Ramesh Natarajan wrote: > We are trying to run TIPC on RHEL 7.2, module compiled using github repo > https://github.com/Hugne/tipc-rhel7.git > > Our setup involves 10 RHEL 7.2 VMs running on 2 ESXi hosts. The ethtool > shows bearer running full duplex ethernet 10000Mb/s. > > Our application sends TIPC messages in a connection oriented mode with all > messages sent using high importance and also subscribe to the topology > service to get notified when a process comes up or down. > > When we run our tests I see a lot of link congestion. I also see dups in > the RX and TX states. > > The immediate question is , is this a known issue and are we missing any > upstream patches on RHEL 7 that could attribute to this behavior? > > It is kind of interesting that our application doesn't seem to be affected > by this congestion. I am not sure if TIPC compensates for the link > congestion some how or we are not testing correctly. > > Can someone explain what is the impact of link congestion? > > Should I be worried about dups in the tipc-config output? > > thanks > Ramesh > > > cat /sys/module/tipc/version > > 2.0.0 > > > uname -a > > Linux pl01.localdomain 3.10.0-327.10.1.el7.x86_64 #1 SMP Sat Jan 23 > 04:54:55 EST 2016 x86_64 x86_64 x86_64 GNU/Linux > > > Link <1.1.1:eth0-1.1.2:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:261417807 fragments:0/0 bundles:56023/166938 > > TX packets:261221765 fragments:0/0 bundles:154215/461415 > > TX profile sample:427258 packets average:316 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:1356713 probes:2484 naks:11272 defs:11576 dups:63802 > > TX states:1079272 probes:2555 naks:15628 acks:1059339 dups:97169 > > Congestion link:3334 Send queue max:80 avg:2 > > > Link <1.1.1:eth0-1.1.3:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:249327535 fragments:0/0 bundles:46132/136797 > > TX packets:249277387 fragments:0/0 bundles:69422/210478 > > TX profile sample:410901 packets average:318 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:1073589 probes:2480 naks:4389 defs:11677 dups:30086 > > TX states:924735 probes:2515 naks:8025 acks:912458 dups:35038 > > Congestion link:1215 Send queue max:80 avg:2 > > > Link <1.1.1:eth0-1.1.4:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:246675382 fragments:0/0 bundles:63742/193772 > > TX packets:246036006 fragments:0/0 bundles:384356/1154001 > > TX profile sample:429353 packets average:319 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:1743106 probes:2484 naks:16703 defs:9825 dups:60508 > > TX states:1148228 probes:2514 naks:18055 acks:1125641 dups:123100 > > Congestion link:9510 Send queue max:80 avg:3 > > > Link <1.1.1:eth0-1.1.5:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:225128324 fragments:0/0 bundles:58476/177973 > > TX packets:224744956 fragments:0/0 bundles:252051/755157 > > TX profile sample:424563 packets average:318 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:1874387 probes:2485 naks:11827 defs:9619 dups:100842 > > TX states:1341783 probes:2518 naks:28955 acks:1308306 dups:87257 > > Congestion link:5564 Send queue max:80 avg:3 > > > Link <1.1.1:eth0-1.1.6:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:221134009 fragments:0/0 bundles:53377/164102 > > TX packets:220738050 fragments:0/0 bundles:255991/762921 > > TX profile sample:422139 packets average:319 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:1826440 probes:2492 naks:11154 defs:8797 dups:101142 > > TX states:1294699 probes:2523 naks:29845 acks:1260272 dups:81031 > > Congestion link:5265 Send queue max:80 avg:3 > > > Link <1.1.1:eth0-1.1.7:eth0> > > ACTIVE MTU:1500 Priority:10 Tolerance:1500 ms Window:50 packets > > RX packets:303149788 fragments:0/0 bundles:167810/579581 > > TX packets:299451019 fragments:0/0 bundles:2166164/6276948 > > TX profile sample:411911 packets average:325 octets > > 0-64:0% -256:0% -1024:100% -4096:0% -16384:0% -32768:0% -66000:0% > > RX states:3865116 probes:2497 naks:25977 defs:10005 dups:56112 > > TX states:3882112 probes:2505 naks:13901 acks:3863833 dups:195330 > > Congestion link:68834 Send queue max:80 avg:3 > ------------------------------------------------------------------------------ > Developer Access Program for Intel Xeon Phi Processors > Access to Intel Xeon Phi processor-based developer platforms. > With one year of Intel Parallel Studio XE. > Training and support from Colfax. > Order your platform today. http://sdm.link/xeonphi > _______________________________________________ > tipc-discussion mailing list > tipc-discussion@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/tipc-discussion ------------------------------------------------------------------------------ Developer Access Program for Intel Xeon Phi Processors Access to Intel Xeon Phi processor-based developer platforms. With one year of Intel Parallel Studio XE. Training and support from Colfax. Order your platform today. http://sdm.link/xeonphi _______________________________________________ tipc-discussion mailing list tipc-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tipc-discussion
