Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.os.linux.networking > #4372 > unrolled thread
| Started by | szr <szr@sREzMOrVEoman.com> |
|---|---|
| First post | 2019-04-26 19:17 -0700 |
| Last post | 2019-04-27 08:03 -0700 |
| Articles | 5 — 4 participants |
Back to article view | Back to comp.os.linux.networking
Quick question about routing szr <szr@sREzMOrVEoman.com> - 2019-04-26 19:17 -0700
Re: Quick question about routing William Unruh <unruh@invalid.ca> - 2019-04-27 03:02 +0000
Re: Quick question about routing Pascal Hambourg <pascal@plouf.fr.eu.org> - 2019-04-27 09:14 +0200
Re: Quick question about routing Theodore Heise <theise@panix.com> - 2019-04-27 14:28 +0000
Re: Quick question about routing szr <szr@sREzMOrVEoman.com> - 2019-04-27 08:03 -0700
| From | szr <szr@sREzMOrVEoman.com> |
|---|---|
| Date | 2019-04-26 19:17 -0700 |
| Subject | Quick question about routing |
| Message-ID | <qa0e4d0egm@news2.newsguy.com> |
Ok, I have a LAN with the following notes: 192.168.1.254 - Main router / default gateway for LAN 192.168.1.4 - Linux server (eth0) 192.168.1.8 - Windows laptop Linux server (192.168.1.4) has ip(v4) forwarding enabled. As an experiment I added an ip, 10.0.0.4, as eth0:0. Then I added a static route for 10.0.0.0/8 using 192.168.1.4 as the gateway on the main router. After that, I ran `ping 10.0.0.4` from the Windows laptop and the ping was successful. While I do have plenty of networking experience and I'm not surprised that this worked, one thing sort of bugs me. I'm able to ping an ip in the 10.0.0.0/8 subnet from a host that only has an ip in the 192.168.1.0/24 subnet. I know that the laptop is using the route from the default gateway (the main router) but how exactly is the ping getting back when the source of the ping isn't coming from another 10.0.0.0/8 address? Is this more due to ip forwarding, or because it's on the same physical network segment? Again, I otherwise have a very good understanding of networking, but I can't seem to resolve this scenario completely in my mind, and would really like to know what exactly is happening here that allows the round trip to occur successfully. -- szr
[toc] | [next] | [standalone]
| From | William Unruh <unruh@invalid.ca> |
|---|---|
| Date | 2019-04-27 03:02 +0000 |
| Message-ID | <qa0go0$s1v$1@dont-email.me> |
| In reply to | #4372 |
On 2019-04-27, szr <szr@sREzMOrVEoman.com> wrote: > Ok, I have a LAN with the following notes: > > 192.168.1.254 - Main router / default gateway for LAN > 192.168.1.4 - Linux server (eth0) > 192.168.1.8 - Windows laptop > > Linux server (192.168.1.4) has ip(v4) forwarding enabled. > > As an experiment I added an ip, 10.0.0.4, as eth0:0. > > Then I added a static route for 10.0.0.0/8 using 192.168.1.4 as the > gateway on the main router. > > After that, I ran `ping 10.0.0.4` from the Windows laptop and the ping > was successful. > > While I do have plenty of networking experience and I'm not surprised > that this worked, one thing sort of bugs me. > > I'm able to ping an ip in the 10.0.0.0/8 subnet from a host that only > has an ip in the 192.168.1.0/24 subnet. I know that the laptop is using > the route from the default gateway (the main router) but how exactly is > the ping getting back when the source of the ping isn't coming from > another 10.0.0.0/8 address? Is this more due to ip forwarding, or > because it's on the same physical network segment? > > Again, I otherwise have a very good understanding of networking, but I > can't seem to resolve this scenario completely in my mind, and would > really like to know what exactly is happening here that allows the round > trip to occur successfully. > Do route -n on the various machines. I suspect that you will have something like 192.168.1.0 0.0.0.0 ... 10.8.0.0 0.0.0.0 ... These say--If you get and address which, when anded with the mask give thos addresses, just send the packet onto the network connected to the machine-- well actually, it sends out a request onto the local network sayins "Who has 10.8.0.5" and the machine that has that address sends back and says "I, who has MAC address 23:55:12:ac:0d, have that address" and your machine sends out a packet addressed to that MAC address.
[toc] | [prev] | [next] | [standalone]
| From | Pascal Hambourg <pascal@plouf.fr.eu.org> |
|---|---|
| Date | 2019-04-27 09:14 +0200 |
| Message-ID | <qa0vg7$7av$1@saria.nerim.net> |
| In reply to | #4372 |
Le 27/04/2019 à 04:17, szr a écrit : > Ok, I have a LAN with the following notes: > > 192.168.1.254 - Main router / default gateway for LAN > 192.168.1.4 - Linux server (eth0) > 192.168.1.8 - Windows laptop > > Linux server (192.168.1.4) has ip(v4) forwarding enabled. Note : IP forwarding is not involved in your scenario. > As an experiment I added an ip, 10.0.0.4, as eth0:0. Note : use of eth0:0 alias is deprecated. You can just add multiple addresses to eth0. > Then I added a static route for 10.0.0.0/8 using 192.168.1.4 as the > gateway on the main router. > > After that, I ran `ping 10.0.0.4` from the Windows laptop and the ping > was successful. > > While I do have plenty of networking experience and I'm not surprised > that this worked, one thing sort of bugs me. > > I'm able to ping an ip in the 10.0.0.0/8 subnet from a host that only > has an ip in the 192.168.1.0/24 subnet. I know that the laptop is using > the route from the default gateway (the main router) No. A routing table is local. The laptop just uses its own routing table. So does any other involved node (host or router) in the path. The laptop routing table has 2 routes : 192.168.1.0/24 direct 0.0.0.0/0 via 192.168.1.254 10.0.0.4 does not match the first route so use the default route and send to the router. The router routing table has 3 routes : 192.168.1.0/24 direct 10.0.0.0/8 via 192.168.1.4 0.0.0.0/0 via some external ISP router 10.0.0.4 matches the second route so forward to 192.168.1.4 (server). [Note : the router may also send an ICMP "redirect" message to the laptop to inform it that 10.0.0.4 is directly reachable on the network segment, and the laptop may update its routing table to reflect this so that it sends subsequent packets to 10.0.0.4 directly to the server.] The server routing table has 3 routes : 192.168.1.0/24 direct 10.0.0.0/8 direct 0.0.0.0/0 via 192.168.1.254 10.0.0.4 is assigned to a local interface so accept the packet and reply to the source 192.168.1.8 (laptop). 192.168.1.8 matches the first route so send directly to the laptop. > but how exactly is > the ping getting back when the source of the ping isn't coming from > another 10.0.0.0/8 address? Is this more due to ip forwarding, or No. As I wrote, no IP forwarding is involved on the server because the destination is the host itself. It is just acting as a host with multiple addresses. > because it's on the same physical network segment? Not really. It would work even if the source was in a distinct network segment. The server would just send the reply back to the router and the router would forward it to the other network segment.
[toc] | [prev] | [next] | [standalone]
| From | Theodore Heise <theise@panix.com> |
|---|---|
| Date | 2019-04-27 14:28 +0000 |
| Message-ID | <slrnqc8poa.71s.theise@panix2.panix.com> |
| In reply to | #4374 |
On Sat, 27 Apr 2019 09:14:14 +0200, Pascal Hambourg <pascal@plouf.fr.eu.org> wrote: > Le 27/04/2019 ?? 04:17, szr a ??crit??: > > Ok, I have a LAN with the following notes: > > > > 192.168.1.254 - Main router / default gateway for LAN > > 192.168.1.4???? - Linux server (eth0) > > 192.168.1.8???? - Windows laptop > > > > Linux server (192.168.1.4) has ip(v4) forwarding enabled. > > Note : IP forwarding is not involved in your scenario. > > > As an experiment I added an ip, 10.0.0.4, as eth0:0. > > Note : use of eth0:0 alias is deprecated. You can just add > multiple addresses to eth0. > > > Then I added a static route for 10.0.0.0/8 using 192.168.1.4 > > as the gateway on the main router. > > > > After that, I ran `ping 10.0.0.4` from the Windows laptop and > > the ping was successful. > > > > While I do have plenty of networking experience and I'm not > > surprised that this worked, one thing sort of bugs me. > > > > I'm able to ping an ip in the 10.0.0.0/8 subnet from a host > > that only has an ip in the 192.168.1.0/24 subnet. I know that > > the laptop is using the route from the default gateway (the > > main router) > > No. A routing table is local. The laptop just uses its own > routing table. So does any other involved node (host or router) > in the path. > > The laptop routing table has 2 routes : > 192.168.1.0/24 direct > 0.0.0.0/0 via 192.168.1.254 > > 10.0.0.4 does not match the first route so use the default route and > send to the router. > > The router routing table has 3 routes : > 192.168.1.0/24 direct > 10.0.0.0/8 via 192.168.1.4 > 0.0.0.0/0 via some external ISP router > > 10.0.0.4 matches the second route so forward to 192.168.1.4 > (server). > > [Note : the router may also send an ICMP "redirect" message to > the laptop to inform it that 10.0.0.4 is directly reachable on > the network segment, and the laptop may update its routing > table to reflect this so that it sends subsequent packets to > 10.0.0.4 directly to the server.] > > The server routing table has 3 routes : > 192.168.1.0/24 direct > 10.0.0.0/8 direct > 0.0.0.0/0 via 192.168.1.254 > > 10.0.0.4 is assigned to a local interface so accept the packet and reply > to the source 192.168.1.8 (laptop). > 192.168.1.8 matches the first route so send directly to the laptop. Very clear explanation, thank you! -- Ted Heise <theise@panix.com> West Lafayette, IN, USA
[toc] | [prev] | [next] | [standalone]
| From | szr <szr@sREzMOrVEoman.com> |
|---|---|
| Date | 2019-04-27 08:03 -0700 |
| Message-ID | <qa1qvi0287n@news3.newsguy.com> |
| In reply to | #4374 |
On 4/27/2019 0:14 AM, Pascal Hambourg wrote: > Le 27/04/2019 à 04:17, szr a écrit : >> Ok, I have a LAN with the following notes: >> >> 192.168.1.254 - Main router / default gateway for LAN >> 192.168.1.4 - Linux server (eth0) >> 192.168.1.8 - Windows laptop >> >> Linux server (192.168.1.4) has ip(v4) forwarding enabled. > > Note : IP forwarding is not involved in your scenario. > >> As an experiment I added an ip, 10.0.0.4, as eth0:0. > > Note : use of eth0:0 alias is deprecated. You can just add multiple > addresses to eth0. > >> Then I added a static route for 10.0.0.0/8 using 192.168.1.4 as the >> gateway on the main router. >> >> After that, I ran `ping 10.0.0.4` from the Windows laptop and the ping >> was successful. >> >> While I do have plenty of networking experience and I'm not surprised >> that this worked, one thing sort of bugs me. >> >> I'm able to ping an ip in the 10.0.0.0/8 subnet from a host that only >> has an ip in the 192.168.1.0/24 subnet. I know that the laptop is >> using the route from the default gateway (the main router) > > No. A routing table is local. The laptop just uses its own routing > table. So does any other involved node (host or router) in the path. > > The laptop routing table has 2 routes : > 192.168.1.0/24 direct > 0.0.0.0/0 via 192.168.1.254 > > 10.0.0.4 does not match the first route so use the default route and > send to the router. > > The router routing table has 3 routes : > 192.168.1.0/24 direct > 10.0.0.0/8 via 192.168.1.4 > 0.0.0.0/0 via some external ISP router > > 10.0.0.4 matches the second route so forward to 192.168.1.4 (server). > > [Note : the router may also send an ICMP "redirect" message to the > laptop to inform it that 10.0.0.4 is directly reachable on the network > segment, and the laptop may update its routing table to reflect this so > that it sends subsequent packets to 10.0.0.4 directly to the server.] > > The server routing table has 3 routes : > 192.168.1.0/24 direct > 10.0.0.0/8 direct > 0.0.0.0/0 via 192.168.1.254 > > 10.0.0.4 is assigned to a local interface so accept the packet and reply > to the source 192.168.1.8 (laptop). > 192.168.1.8 matches the first route so send directly to the laptop. > >> but how exactly is the ping getting back when the source of the ping >> isn't coming from another 10.0.0.0/8 address? Is this more due to ip >> forwarding, or > > No. As I wrote, no IP forwarding is involved on the server because the > destination is the host itself. It is just acting as a host with > multiple addresses. > >> because it's on the same physical network segment? > > Not really. It would work even if the source was in a distinct network > segment. The server would just send the reply back to the router and the > router would forward it to the other network segment. Thanks, this all makes complete sense now. I must have been a bit tired as my mind was somehow crossing up this with trying to reach hosts behind a masqueraded ip, vlan, or some similar scenario. Just one of those cases where after a good nights rest it all becomes clear. Thanks again. -- szr
[toc] | [prev] | [standalone]
Back to top | Article view | comp.os.linux.networking
csiph-web