Re: [TECH] academique, routage BGP asymetrique casse TLS 443

jehan procaccia INT via frnog <[email protected]>
Newsgroups gmane.org.operators.frnog
Organization IMT
Message-ID <[email protected]>
Ok, dernier jet avant le weekend avec cette capture des 2 cotés pour 
avoir  un os a ronger ;-)

coté "client" VPS OVH qui initie la connexion

# openssl s_client -connect srv.int-evry.fr:443
Connecting to 157.150.11.6
CONNECTED(00000003)
write:errno=104

ce meme client voit ceci en filtrant sur l'IP du serveur sollicité

/# tcpdump -i eth0 host 157.150.11.6 and 'tcp[tcpflags] & tcp-syn != 0' 
-nnvv
dropped privs to tcpdump
tcpdump: listening on eth0, link-type EN10MB (Ethernet), snapshot length 
262144 bytes


16:52:54.044825 IP (tos 0x0, ttl 64, id 53493, offset 0, flags [DF], 
proto TCP (6), length 60)
     51.178.25.200.46744 > 157.150.11.6.443: Flags [S], cksum 0xf58d 
(incorrect -> 0xf8d3), seq 1392587108, win 65320, options [mss 
1420,sackOK,TS val 3947115214 ecr 0,nop,wscale 7], length 0
16:52:54.050731 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x7651 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467895586 ecr 3947115214,nop,wscale 8], length 0
16:52:55.452651 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x70d7 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467896988 ecr 3947115214,nop,wscale 8], length 0
16:52:57.452632 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x6907 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467898988 ecr 3947115214,nop,wscale 8], length 0
16:53:01.452679 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x5967 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467902988 ecr 3947115214,nop,wscale 8], length 0
16:53:09.452810 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x3a27 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467910988 ecr 3947115214,nop,wscale 8], length 0

16:53:25.452647 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xfba6 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467926988 ecr 3947115214,nop,wscale 8], length 0

^C
7 packets captured
9 packets received by filter
0 packets dropped by kernel/

J'ai pourtant l'impression qu'il reçoit bien les reponses dès le 2eme 
packet

/16:52:54.050731 IP (tos 0x0, ttl 50, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0x7651 
(correct), seq 1591607196, ack 1392587109, win 28960, options [mss 
1460,sackOK,TS val 3467895586 ecr 3947115214,nop,wscale 8], length 0/

Qui correspond a la reponse coté serveur

/17:52:54.047975 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x7651), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467895586 ecr 3947115214,nop,wscale 8], 
length 0/

=> meme /seq 1591607196,/ qui ack /1392587109 => /demande initiale 
en seq/1392587108 ,/ bref tout semble bon jusque là , non ?

Coté Serveur tj la meme chose au total

/# tcpdump -i eth0 host 51.178.25.200 and 'tcp[tcpflags] & tcp-syn != 0' 
-nnvv
dropped privs to tcpdump
tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 
262144 bytes


17:52:54.047928 IP (tos 0x0, ttl 48, id 53493, offset 0, flags [DF], 
proto TCP (6), length 60)
     51.178.25.200.46744 > 157.150.11.6.443: Flags [S], cksum 0xf8d3 
(correct), seq 1392587108, win 65320, options [mss 1420,sackOK,TS val 
3947115214 ecr 0,nop,wscale 7], length 0
17:52:54.047975 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x7651), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467895586 ecr 3947115214,nop,wscale 8], 
length 0
17:52:55.449927 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x70d7), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467896988 ecr 3947115214,nop,wscale 8], 
length 0
17:52:57.449883 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x6907), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467898988 ecr 3947115214,nop,wscale 8], 
length 0
17:53:01.449928 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x5967), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467902988 ecr 3947115214,nop,wscale 8], 
length 0
17:53:09.449913 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0x3a27), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467910988 ecr 3947115214,nop,wscale 8], 
length 0

17:53:25.449930 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto 
TCP (6), length 60)
     157.150.11.6.443 > 51.178.25.200.46744: Flags [S.], cksum 0xf58d 
(incorrect -> 0xfba6), seq 1591607196, ack 1392587109, win 28960, 
options [mss 1460,sackOK,TS val 3467926988 ecr 3947115214,nop,wscale 8], 
length 0

^C
7 packets captured
7 packets received by filter
0 packets dropped by kernel/

Bref je reste perplexe ...

Merci


On 3/6/26 17:18, David Ponzone wrote:
> Faut prendre une trace tcpdump sur le client aussi, pour comparer 
> paquet par paquet, ça te dira s’il y a des paquets retour qui ont été 
> filtrés.
> Après, on peut faire une comparaison des 2 traces avec Wireshark à 
> priori (sujet abordé dans un autre fil récemment).
>
> David
>
>> Le 6 mars 2026 à 16:15, jehan procaccia INT 
>> <[email protected]> a écrit :
>>
>> MErci , mais bof .. cela ne change pas apparement
>>
>> sur le client VPS OVH je fait
>>
>> # ip link set dev eth0 mtu 1460
>>
>> c'est bien en place
>>
>> 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu *1460* qdisc fq_codel 
>> state UP group default qlen 1000
>>
>> je lance mon test
>>
>> # openssl s_client -connect srv.int-evry.fr:443
>>
>> meme echec
>>
>> coté server destination je vois bien arriver la requetes comme avant 
>> avec un Syn from /51.178.25.200 (VPS OVH) /et des reponses Syn-Ack 
>> depuis mon serveur lcoal (/157.150.11.6) /qui ne semblent pas revenir 
>> sur le VPS OVH :-(
>>
>> # tcpdump -i eth0 host 51.178.25.200 and  'tcp[tcpflags] & tcp-syn != 
>> 0' -nnvv
>>
>> /17:08:01.589820 IP (tos 0x0, ttl 48, id 2470, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     51.178.25.200.40236 > 157.150.11.6.443: Flags [S], cksum 0xd098 
>> (correct), seq 412271373, win 65320, options [mss 1420,sackOK,TS val 
>> 3944422756 ecr 0,nop,wscale 7], length 0
>> 17:08:01.589873 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0xba7f), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465203128 ecr 3944422756,nop,wscale 
>> 8], length 0
>> 17:08:02.591909 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0xb695), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465204130 ecr 3944422756,nop,wscale 
>> 8], length 0
>> 17:08:04.591924 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0xaec5), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465206130 ecr 3944422756,nop,wscale 
>> 8], length 0
>> 17:08:08.591929 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0x9f25), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465210130 ecr 3944422756,nop,wscale 
>> 8], length 0
>> 17:08:16.793894 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0x7f1b), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465218332 ecr 3944422756,nop,wscale 
>> 8], length 0
>> 17:08:32.794923 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], 
>> proto TCP (6), length 60)
>>     157.150.11.6.443 > 51.178.25.200.40236: Flags [S.], cksum 0xf58d 
>> (incorrect -> 0x409a), seq 466676691, ack 412271374, win 28960, 
>> options [mss 1460,sackOK,TS val 3465234333 ecr 3944422756,nop,wscale 
>> 8], length 0
>> ^C
>> 7 packets captured
>> 7 packets received by filter
>> 0 packets dropped by kernel/
>>
>>
>>
>> On 3/6/26 16:38, David Ponzone wrote:
>>> Ou en forçant le TCP MSS à 1420 par exemple.
>>>
>>> David Ponzone
>>>
>>>
>>>
>>>> Le 6 mars 2026 à 15:34, P. Deloney via frnog<[email protected]> a écrit :
>>>>
>>>> jehan procaccia INT via frnog<[email protected]>:
>>>>
>>>>> oui j'ai des traces [1]
>>>> Ça ressemble clairement à un problème de MTU ou une mauvaise
>>>> configuration d'un pare-feu "stateful".
>>>>
>>>> Je ferais des tests en baissant le MTU de l'interface de l'hôte qui
>>>> initie la connexion TCP.
>>>>
>>>>
>>>> ---------------------------
>>>> Liste de diffusion du FRnOG
>>>> http://www.frnog.org/
>
---------------------------
Liste de diffusion du FRnOG
http://www.frnog.org/
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.