packet loss on loopback interface!?

  • Permalink
  • submit to reddit
  • Email
  • Follow

I am using udp as communication protocol. I know that udp is connection 
less and may lose datagrams, but i expected this only on real networks 
or buffer overflows, not when using the loopback interface.

My application performs an handshaking over udp, the client sends an 
request and the server ansers with an ack. Fairly simple, but when 
running on max speed with no wait states, some datagrams are lost.
But why? Sending request-response is a form of flow-control, so the 
network buffers for udp should not overflow.

Is thery any way to reduce the discardment of datagrams, possibly a 
socket option or solaris kernel parameters? The problem occurs on real 
network connections (not loopback), too.
I used the option SO_DEBUG with my socket, but it seems there's no 
debug-log written to the syslog.
Are there any other tuning parameters!?

regards, markus

Reply Markus 1/15/2004 8:12:47 AM

See related articles to this posting

Markus Breuer <> writes:

>My application performs an handshaking over udp, the client sends an 
>request and the server ansers with an ack. Fairly simple, but when 
>running on max speed with no wait states, some datagrams are lost.
>But why? Sending request-response is a form of flow-control, so the 
>network buffers for udp should not overflow.

If you do request response I would not expecting packetloss; if you
are sending as fast as you can and you're not waiting for responses
then the kernel will not flow control UDP buffers will overflow
and packets will be dropped.

>Is thery any way to reduce the discardment of datagrams, possibly a 
>socket option or solaris kernel parameters? The problem occurs on real 
>network connections (not loopback), too.
>I used the option SO_DEBUG with my socket, but it seems there's no 
>debug-log written to the syslog.

You can increase the udp buffer size using setsockopt() and the
max buffer size using ndd -set /dev/udp.

For loopback, you can use AF_UNIX, SOCK_SEQPACKET

Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.
Reply Casper 1/15/2004 8:43:00 AM
comp.unix.solaris 25854 articles. 89 followers. Post

1 Replies

Similar Articles

[PageSpeed] 47

  • Permalink
  • submit to reddit
  • Email
  • Follow


Similar Artilces:

Packet loss with libpcap on a loopback interface
Hi! I'm trying to write a protocol decoder using libpcap. I'm having problems with a simple test program that sends a simple "hello world"- string on a TCP port. I intermittently receive some "TCP ACKed lost segment" for the packets that are sent/received. My best guess is that this is related to either libpcap or my setup since I get the same behaviour using my own protocol decoder as with Wireshark. The fact that the segments are not lost, other than for libpcap, are obvious since the test application receives the data without any retransmissions going ...

Packet loss on larger packets only
Being stumped on a problem with a Win2K (SP4) box. Recently started to have network performance problems, no recent configuration changes that I can identify that might be responsible. I've narrowed the problem down to an issue where small packets are OK, but I get lots of packet loss on packets larger than 200-250 bytes, particularly over 300 bytes. This was tested by using "ping -t -l <size> <destination>" for hosts either on the same LAN or over an internet link. Problem manifests itself whether pinging from or to the affected host. This box also h...

packet loss because of big packets
Hello, The following problem occurs in my network: The command 'ping anyhost' works well without packet loss. Ping sends packets with a size of about 65 Bytes. If I grow the packet size with the parameter -s, I have got packet loss. The command 'ping -s 65507 host' works well for some hosts, for others not (100 % loss). Does this indicate bad network components? What does this mean for other protocols like UDP or TCP? Thank You for Your help! Andreas This might be due to certain hosts not supporting ICMP Echo reply with larger packet sizes. Sandeep -- htt...

New packet loss issue
I've submitted this as a Sun case, but I thought I'd post here as well to see if anyone else has experienced this problem. Any ideas? ------------ The server is an x4100 M2 running Solaris 10 11/06 x86. Patches were current as of 4/9/2007, including a reconfig reboot at that time. Yesterday, the server was rebooted and afterwards the packet loss was noticed immediately. Note that as part of troubleshooting this problem, I've brought the patches up to date as of yesterday, performed a reconfig reboot, as well as performing a cold boot (with power cables completely removed). ...

Reapeatable packet loss on ulip interface over Solaris TCP on 100Mbps
Hi I use VxSim with ulip over Solaris8. Whenever the gap between 2 packets < 10usec - The second received packet does not arrive to ulip interface. This cause a very slow TCP connection. On 10BaseT it never happens (probably because the inter packet gap > 10 usec). Did anyone encountered this problem? Thanks, Ilan ...

LoopBack Interface
Can anybody help me in the following subject? Situation 1. Router 1 (2500 series) have 2 Serial Interface connected to 2 Branches. Loopback Interface is also configured for the Router & is running OSPF as the Routing Protocol. Does the Loopback Interface have any role to play if both the Serial Interfaces goes down(Line Protocol)? Thanks.. In article <>, Sabya <> wrote: :Can anybody help me in the following subject? :Situation 1. Router 1 (2500 series) have 2 Serial Interface connected :to 2 Branche...

Loopback Interface
I'm running a slave kerberos KDC on my OS X laptop. I need to access the KDC when the laptop is disconnected from any network, but the KDC doesn't bind to the loopback interface, and I can't seem to setup a dummy interface in OS X. OS X shuts down any interface on which it can't detect a link. How can I force the KDC to bind to the loopback interface? Thanks, Jack ________________________________________________ Kerberos mailing list >>>>> "ms419" == ms419 <ms419@...

interface to interface
Can I write a program to read packet from one interface and output packet to another interface? I wonder if os identify the two IP belong to interface in local machine, transmitting the packet in IP layer,and no packet goes out the interface. Is it possible to happen? In article <>, (Joseph) wrote: > Can I write a program to read packet from one interface and output > packet to another interface? If you can write a program to read from one interface, and you can write a program to write to the other interface, wh...

loopback interface
Is it possible to create many loopback interfaces (as in Cisco routers) and get different statistics for these interfaces ? ...

Packet loss?
Hi, As an exercise I am trying to implement a very simple packet-loss measurement tool (copying the approach in Sting: a TCP-based Network Measurement Tool by Stefan Savage). I am sending a padded HTTP GET request as a sequence of TCP packets with 1 byte of payload to various web servers and watching the acks coming back. I pause for 1 second between sending each TCP packet. I have two questions: 1. I am not seeing any packet loss. Am I getting lucky or is the size of the packet (1 byte payload) somehow not big enough to encourage packet loss? 2. Some web servers are happy ...

Packet loss
Hello, Please, could anyone of you tell me why packet loss happens in a computer network? Need to know when and why a packet is rejected and disappears, and what takes that decision. I really need a deep explanation or a source where I can dig in. Being quite new in this subject, I have not found a clear explanation to this problem. Thanks a lot. -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 LMMT wrote: | Hello, | | Please, could anyone of you tell me why packet loss happens in a | computer network? Sunspot activity, backhoes, software bugs, firewalls, oxydation on conn...

problem with ethernet loopback test using loopback plug/stub: receiving socket not seeing packet.
we are writing socket application to send UDP packets out from eth1 and loop it back to the same interface using loopback stub to verify the hardware without using external devices/ports. The sending and receiving port are the same. our target machine has eth0 configured as 15.6.xx.xx and eth1 configured as the listening socket is listing on all ip. .. When we looked at the packet in the tx code of the driver, we have observed that the packet passed down from the ip stack has the same MAC address for both source and destination. Also when packet is returning, its ip source and...

loopback interface / VPN
I set up my router 877 to use EzVPN/xauth to connect to the office router. Whenever it succeeds in making a VPN connection ( for testing, I do it many times ) the config shows another line of interface loopbackN. Right now it shows : interface Loopback0 ip address ! interface Loopback1 ip address ! interface Loopback2 ip address ! interface Loopback3 ip address ! Can this cause any problem if I just let them there ? Or do I have to clear the old ones ? In the...

Network packet loss
I'm having trouble with a K570 I've got in my garage. I want to access the K box from a PC inside the house. So I came up with the idea to link the systems via the wireless network I have in the house, a 2Wire router/modem -1000HG, via SBC. I got a Motorola bridge, WE800G and setup the wireless connection successfully. I�m able to ping the bridge from a PC in the house, no problem. Here�s the problem, while I can ping the bridge with 0 packet loss, if I try to ping the 9000 (which is Cat 5 Ethernet connected to the bridge) I get a typically 75% failure (only 4 tries from ...

Winsock + Loopback Interface
Hi, While addressing a bug in the implementation of Tcl sockets under Windows: I'm observing strange things only on the Windows loopback interface. These strange things seem to be related to TCP FIN packets. However, the only tool I've found so far to snoop the loopback is CommView, and it seems to be unable to see the FIN flag on the loopback (even for a bug-free, normally terminated socket). FINs are correctly reported on non-loopback. Question 1: do you know of another tool, ...

some commands about loss packet
Hi On the c 4500 is command: show platform show platform cpu packet statistics all Now i have c 4006 and i can't find this or similar command. I'd like to now about loss packet when the cpu is overhead or near overhead and when the proces packet forwardig is big how many packets is droped. The command sh proc cpu is not enough, i think... regards popo If you are running CEF, then see sh...

50% packet loss ?
Hi. What does it meen when my ping reply is !.!.!.!.!.!.!. for one IP-address. In LAN the ping reply is !!!!!!!! but from another network only every other packet is comming back. Kenneth, Check for static route to non-existing or dead gateway. For example, if you have two static routes with the same metrics to the same destination, router will try to do a load-balancing between gateways, and if one is down, you will get every other packet lost. Good luck, Mike CCNP, CCDP, CCSP, Cisco Voice, MCSE W2K, MCSE+I, Security+, etc. CCIE R&S (in progress), CCIE Voice (in progress) -----...

TCP_NODELAY at loopback interface
Hello, the WINCE 5.0 Winsock socket option TCP_NODELAY should control, if the Nagle algorithm is activated or not. The nagle alrorithm should delay transmission of short segments and send one segments, that contains all the short chunks. see 1) Does the Nagel algorithm also works for the Loopback interface How is the term "data already in flight" to be read? 2) The function getsockopt returns the current value of the winsock option. What does it mean, when I get the result 8 (eight), when I have set the v...

Freeswan and loopback interface
Hi all! I'm trying to solve without success the following problem: Scenario: I've got a linux appliance behind an adsl router. I user a private /30 to connect the router to the appliance to save IP addresses. The router does only routing no nat. | lo:1 public ip1 int adsl: public ip2 |----Linux appliance---------------ADSL router-------- | eth0: eth: private net In this way I use only 2 pubblic IP addresses. Don't tell me to use an ADSL interface on the appliance: I can't. Problem: I want to set up a IPSec ...

System Packet Interface?
Hi, I am doing a project on a 10GB Ethernet. I am going to use a FPGA with a 10Gb Ethernet Mac Core. If I use a NIOS II, it's not necessary for me to purchase a developmen board supporting the SPI (System Packet Interface)? This interface is used between a FPGA and NPU. Thank you On Fri, 24 Feb 2006 09:11:02 -0600, "freechip" <> wrote: > >Hi, >I am doing a project on a 10GB Ethernet. >I am going to use a FPGA with a 10Gb Ethernet Mac Core. >If I use a NIOS II, it's not necessary for me to purchase a development >board supporting...

Multiplexing and packet loss
Hi I have a 1gbit (1000BaseZX) link between two sites that's, due to large number of small packets, have alot of buffer related packet loss in one direction. A solution I'm planning to implement is CWDM on this link and team at least four interfaces into an etherchannel. I'm however unsure if this will actually solve any of the packer loss problems. In theory four interface buffers should be better than one but I wounder if anyone have any real world data on this. Will a CWDM etherchannel only move the buffer problems from the interface to the etherchannel buffer? Regards Fredrik...

6506 loopback interface
We have a single 6506 which we will replace with two 6506 in HSRP redundancy mode. Until now we don't have a loopback interface configured on the 6506. Most management articles advise the use of the loopback interface so we want to configure it on both the 6506's. What we can't find, is what kind of address to put on both machines. Loopback addresses seem to be host adresses but how are they related to each other ? If we give router A address x.x.y.1, can we give router B address x.x.y.2 or do we need to give it x.x.Z.1 so i need a separate subnet for each interface ? If one...

1841 Packet loss
Here is the situation, 1841 with connected 100Mbs LAN and 100Mbs WAN (capped at 20Mbs). It's used to send VOIP calls. THe LAN side is connected to a VOIP GW sending via calls 16T1s, so 368 simulataneous calls. When all 16 T1s are used we gets packet loss (between point to point's on the WAN link) and reports of garbled calls, if we only use 8 T1s all is well. Here is a snapshot of the Traffic on the links.. 30 second input rate 2003000 bits/sec, 2734 packets/sec 30 second output rate 3545000 bits/sec, 5000 packets/sec So including both FE's that about 16000pps... lets 20kp...

Redirect packets between interfaces.
Hej, mam sobie taki problem, ze chce, aby czesc ruchu (glownie P2P) rutowana byla przez interfejs ppp1. "Glowniejszy" defaultroute jest na ppp0, na ppp1 nie dziala traceroute (ktos ma pomysl, czemu?), ale polaczyc sie da, ustawiajac recznie regulke ip route dla siakiegos ip. Hi, I have the following problem. I read iptables tutorials, LARTC, without any clue for solution. Anyway, I have two interfaces: tun0 is reliable, fast, but has most P2P blocked on the other side ppp0 is slower, but is totally non-firewalled. I would like to use L7 to classify packets as P2P traffic and then ...