<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Hi,<br>
    <br>
    PROTO=ICMP TYPE=11 CODE=0 <br>
    <br>
    <span style="color: rgb(34, 34, 34); font-family: sans-serif;
      font-size: 14px; font-style: normal; font-variant-ligatures:
      normal; font-variant-caps: normal; font-weight: 400;
      letter-spacing: normal; orphans: 2; text-align: start;
      text-indent: 0px; text-transform: none; white-space: normal;
      widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px;
      background-color: rgb(248, 249, 250); text-decoration-style:
      initial; text-decoration-color: initial; display: inline
      !important; float: none;">TTL expired in transit</span><br>
    <br>
    perhaps a routing loop? <br>
    <br>
    regards<br>
    <br>
    Thomas <br>
    <br>
    <div class="moz-cite-prefix">Am 30.12.17 um 22:47 schrieb Martin
      Sand:<br>
    </div>
    <blockquote cite="mid:f94f553d-df25-0a39-de92-2d4c26cf11ea@gmx.net"
      type="cite">Hi Noel
      <br>
      <br>
      Thanks for the advice. I installed tcpdump and wireshark and added
      a rule to log ICMP errors.
      <br>
      This is an excerpt from the log file. I assume this line shows
      something is sent to port 80 but I cannot find the corresponding
      iptables entry.
      <br>
      <br>
      Dec 30 21:42:11 localhost kernel: [1423944.393321] IN= OUT=eth0
      SRC=210.211.212.213 DST=192.168.2.135 LEN=88 TOS=0x00 PREC=0xC0
      TTL=64 ID=38805 PROTO=ICMP TYPE=11 CODE=0 [SRC=192.168.2.135
      DST=192.168.1.130 LEN=60 TOS=0x00 PREC=0x00 TTL=1 ID=63979 DF
      PROTO=TCP SPT=47511 DPT=80 WINDOW=5840 RES=0x00 SYN URGP=0 ]
      <br>
      <br>
      Best regards
      <br>
      Martin
      <br>
      <br>
      <br>
      On 28.12.2017 01:43, Noel Kuntze wrote:
      <br>
      <blockquote type="cite">Hi,
        <br>
        <br>
        Looks like your firewall rules on the hub are broken and cause
        the problems or you need to configure an additional CHILD_SA to
        tunnel ICMP errors from the hub, because it has no IP in the
        local TS.
        <br>
        Check both those suspicions.
        <br>
        <br>
        Kind regards
        <br>
        <br>
        Noel
        <br>
        <br>
        On 27.12.2017 23:00, Martin Sand wrote:
        <br>
        <blockquote type="cite">Thanks again Noel.
          <br>
          <br>
          I have executed `traceroute -T --mtu <destination>` and
          `mtr -rw <destination>` on machines at both locations.
          <br>
          I did not do further investigation on the MSS yet since I have
          this strange packet loss.
          <br>
          Based on the route, I assume this happens at the hub which is
          in between the two routers?
          <br>
          Could this be the root cause I need to further investigate?
          <br>
          <br>
          Kind regards
          <br>
          Martin
          <br>
          <br>
          traceroute -T --mtu pi-frankfurt
          <br>
          traceroute to pi-frankfurt (192.168.2.135), 30 hops max, 60
          byte packets
          <br>
            1  router-freiburg (192.168.1.1)  0.263 ms  0.179 ms  0.172
          ms
          <br>
            2  * * *
          <br>
            3  router-frankfurt (192.168.2.1)  41.762 ms  41.182 ms 
          36.716 ms
          <br>
            4  pi-frankfurt (192.168.2.135)  36.693 ms  43.629 ms 
          37.051 ms
          <br>
          <br>
          traceroute -T --mtu pi-freiburg
          <br>
          traceroute to pi-freiburg (192.168.1.130), 30 hops max, 60
          byte packets
          <br>
            1  router-frankfurt (192.168.2.1)  0.489 ms  0.381 ms  0.287
          ms
          <br>
            2  * * *
          <br>
            3  router-freiburg (192.168.1.1)  38.368 ms  47.673 ms 
          35.441 ms
          <br>
            4  pi-freiburg (192.168.1.130)  39.456 ms  54.566 ms  36.117
          ms
          <br>
          <br>
          mtr -rw pi-frankfurt
          <br>
          Start: 2017-12-27T22:57:40+0100
          <br>
          HOST: workstation              Loss%   Snt   Last   Avg  Best
          Wrst StDev
          <br>
             1.|-- router-freiburg         0.0%    10    0.2   0.2   0.2
          0.3   0.0
          <br>
             2.|-- ???                      100.0    10    0.0   0.0  
          0.0 0.0   0.0
          <br>
             3.|-- router-frankfurt        0.0%    10   33.3  35.5  32.5
          42.0   2.7
          <br>
             4.|-- pi-frankfurt              0.0%    10   33.5  34.4 
          32.7 36.7   1.5
          <br>
          <br>
          <br>
          On 27.12.2017 21:08, Noel Kuntze wrote:
          <br>
          <blockquote type="cite">Hi,
            <br>
            <br>
            You can test the convergence speed using `traceroute -T
            --mtu <destination>`, but that only gives you the MTU.
            You need to manually discover the MSS
            <br>
            using `traceroute -T -O mss=<mss>
            <destination>`.
            <br>
            <br>
            The best way to check if the problem continues is to just
            run tcpdump/wireshark and check for ICMP Fragmenation needed
            packets and TCP errors or timeouts.
            <br>
            <br>
            Kind regards
            <br>
            <br>
            Noel
            <br>
            <br>
            On 27.12.2017 17:12, Martin Sand wrote:
            <br>
            <blockquote type="cite">Thanks Noel. Sorry, I had to travel
              to the other location (350 km).
              <br>
              <br>
              I adapted the iptable rules. It improved, but I have the
              impression it only improved a bit.
              <br>
              Is there a way to measure MTU discovery time?
              <br>
              <br>
              Kind regards
              <br>
              Martin
              <br>
              <br>
              <br>
              On 14.12.2017 13:51, Noel Kuntze wrote:
              <br>
              <blockquote type="cite">Hi,
                <br>
                <br>
                <blockquote type="cite">VPN internal http requests to a
                  web server of another spoke take some time until the
                  page is rendered.
                  <br>
                  I assume this is due to the latency.
                  <br>
                </blockquote>
                Nah. It's extremely more likely that the path MTU
                discovery takes some time (maybe due to some
                missing/wrong firewall rules on some host(s) in your
                network topology).
                <br>
                Try lowering the MTU and MSS of the tunneled traffic[1].
                <br>
                <br>
                Kind regards
                <br>
                <br>
                Noel
                <br>
                <br>
                [1]
<a class="moz-txt-link-freetext" href="https://wiki.strongswan.org/projects/strongswan/wiki/ForwardingAndSplitTunneling#MTUMSS-issues">https://wiki.strongswan.org/projects/strongswan/wiki/ForwardingAndSplitTunneling#MTUMSS-issues</a><br>
                <br>
                On 14.12.2017 13:41, Martin Sand wrote:
                <br>
                <blockquote type="cite">Hi all
                  <br>
                  <br>
                  I have a Hub and Spoke setup. Connections are working
                  perfectly fine.
                  <br>
                  Throughput is almost reaching the maximum rate of the
                  upload channel speed, 10 MBit/s.
                  <br>
                  <br>
                  Unfortunately the latency is not fulfilling my
                  objectives. I have an average ping time of 39 ms (see
                  below) when pinging clients on other spokes.
                  <br>
                  VPN internal http requests to a web server of another
                  spoke take some time until the page is rendered.
                  <br>
                  I assume this is due to the latency.
                  <br>
                  <br>
                  Is there any chance to improve the latency? Or is the
                  latency perfectly good?
                  <br>
                  <br>
                  Best regards
                  <br>
                  Martin
                  <br>
                  <br>
                  Hub internet address
                  <br>
                  64 bytes from vpn.example.com (217.122.5.6):
                  icmp_seq=1 ttl=57 time=15.2 ms
                  <br>
                  <br>
                  Internal address of Hub
                  <br>
                  PING 192.168.1.1 (192.168.1.1) 56(84) bytes of data.
                  <br>
                  64 bytes from 192.168.1.1: icmp_seq=1 ttl=62 time=40.4
                  ms
                  <br>
                  <br>
                  Client on another spoke
                  <br>
                  PING 192.168.1.130 (192.168.1.130) 56(84) bytes of
                  data.
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=1 ttl=61
                  time=108 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=2 ttl=61
                  time=41.8 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=3 ttl=61
                  time=38.0 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=4 ttl=61
                  time=35.2 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=5 ttl=61
                  time=36.4 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=6 ttl=61
                  time=39.1 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=7 ttl=61
                  time=38.1 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=8 ttl=61
                  time=41.6 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=9 ttl=61
                  time=36.0 ms
                  <br>
                  64 bytes from 192.168.1.130: icmp_seq=10 ttl=61
                  time=36.7 ms
                  <br>
                  <br>
                  --- 192.168.1.130 ping statistics ---
                  <br>
                  10 packets transmitted, 10 received, 0% packet loss,
                  time 9013ms
                  <br>
                  rtt min/avg/max/mdev = 35.295/45.159/108.281/21.146 ms
                  <br>
                  <br>
                </blockquote>
              </blockquote>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <br>
    </blockquote>
    <br>
    <pre class="moz-signature" cols="72">-- 
Thomas Will

Xinux e.K.
Wichernstrasse 18
66482 Zweibruecken

Registergericht
Amtsgericht Zweibruecken
HRA 1518

P: +49 6332 44040
F: +49 6332 899227
M: +49 170 5218548
M: +49 176 97497102

E: <a class="moz-txt-link-abbreviated" href="mailto:thomas.will@xinux.de">thomas.will@xinux.de</a>
W: <a class="moz-txt-link-freetext" href="http://www.xinux.de">http://www.xinux.de</a>
</pre>
  </body>
</html>