RedHat client ignores DUP ACKs and waits 200ms to retransmit
Linux - NetworkingThis forum is for any issue related to networks or networking.
Routing, network cards, OSI, etc. Anything is fair game.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
RedHat client ignores DUP ACKs and waits 200ms to retransmit
Mu understanding of RFC 1122 (Fast Retransmit) is that upon
receipt of the 3rd duplicate ACK, the sender should react
by starting to retransmit the lost packets.
In my case, looking at the Wireshark trace, I see the sender receiving hundreds of DUP ACKS from the receiver and doing nothing about it until 200 ms pass by. Only then does the
sender finally starts to retransmit. That's killing performance.
The sender is running Red Hat 5.3 (which is Linux 2.6.18).
Is this an indication that the sender does not support
Fast Retransmit? Is there a config param that I need to turn
on?
This seems pretty pertinent to you... http://kerneltrap.com/mailarchive/li...0/6/30/6280312 the official patch they discuss seems to have gone into 2.6.19 but I would've expected Redhat to have back ported it to the more recent el5 kernels through rhn.
Hi, I have a similar problem on RHEL5.3.
Client(Windows) sends many duplicate acks to server(RHEL5.3), but server doesn't retransmit the lost packet. I think this is same problem.
Do you have any latest information with this problem?
The packet trace is the following. #24's seq is 2844946993, which was lost packets, and client is requesting this packets many times.
20 15:40:16.799955 server client TCP 1514 [TCP segment of a reassembled PDU]
21 15:40:16.799958 server client TCP 1514 [TCP segment of a reassembled PDU]
22 15:40:16.799960 server client TCP 1514 [TCP segment of a reassembled PDU]
23 15:40:16.805769 client server TCP 66 16578 > 443 [ACK] Seq=1285985661 Ack=2844942649 Win=23168 Len=0 TSval=2523566062 TSecr=2081593646
24 15:40:16.805853 server client TCP 1514 [TCP segment of a reassembled PDU]
25 15:40:16.805856 server client TCP 1514 [TCP segment of a reassembled PDU]
26 15:40:16.805861 server client TLSv1 1514 Application Data
27 15:40:16.807041 client server TCP 66 16578 > 443 [ACK] Seq=1285985661 Ack=2844945545 Win=28960 Len=0 TSval=2523566062 TSecr=2081593652
28 15:40:16.807195 server client TCP 1514 [TCP segment of a reassembled PDU]
29 15:40:16.813729 client server TCP 66 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566063 TSecr=2081593652
30 15:40:16.813815 server client TCP 1514 [TCP segment of a reassembled PDU]
31 15:40:16.813817 server client TCP 1514 [TCP segment of a reassembled PDU]
32 15:40:16.813886 server client TCP 1514 [TCP segment of a reassembled PDU]
33 15:40:16.813889 server client TLSv1 1514 Application Data
34 15:40:16.814989 client server TCP 66 [TCP Dup ACK 29#1] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566063
35 15:40:16.815069 server client TCP 1514 [TCP segment of a reassembled PDU]
36 15:40:16.819633 client server TCP 66 [TCP Dup ACK 29#2] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
37 15:40:16.819714 server client TLSv1 1514 Ignored Unknown Record
38 15:40:16.819790 client server TCP 66 [TCP Dup ACK 29#3] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
39 15:40:16.819792 client server TCP 66 [TCP Dup ACK 29#4] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
40 15:40:16.820063 server client TLSv1 1514 Ignored Unknown Record
41 15:40:16.820066 server client TLSv1 1514 Ignored Unknown Record
42 15:40:16.820075 client server TCP 66 [TCP Dup ACK 29#5] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
43 15:40:16.820694 client server TCP 66 [TCP Dup ACK 29#6] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
44 15:40:16.824938 client server TCP 66 [TCP Dup ACK 29#7] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
45 15:40:16.825369 client server TCP 66 [TCP Dup ACK 29#8] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
46 15:40:16.825628 client server TCP 66 [TCP Dup ACK 29#9] 16578 > 443 [ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523566064
47 15:41:46.795466 client server TCP 66 16578 > 443 [FIN, ACK] Seq=1285985661 Ack=2844946993 Win=31856 Len=0 TSval=2523575062 TSecr=2081593
48 15:41:46.835782 server client TCP 66 443 > 16578 [ACK] Seq=2844964369 Ack=1285985662 Win=8330 Len=0 TSval=2081683685 TSecr=2523575062
49 15:42:16.817706 server client TCP 1514 [TCP Retransmission] [TCP segment of a reassembled PDU]
50 15:42:16.823511 client server TCP 60 16578 > 443 [RST] Seq=1285985662 Win=0 Len=0
Yes, your suggestion is right and I think so, too.
But this problem happened on my customer business system, so I have to explain which kernel fixed and tell him detail information.
Is it correct that kernel 2.6.19 is fixed this problem?
I serched in redhat knowledgebase and bugzilla but I couldn't find any information of this problem.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.