0

Background:

I have an isolated LAN. The switch is configured with 3 VLANs Here is how my network is setup

VLAN   Subnet           Mask            CIDR    Range               Gateway
10     192.168.10.0     255.255.255.224 /27     192.168.10.1-32     192.168.10.1
15     192.168.10.96    255.255.255.224 /27     192.168.10.97-126   192.168.10.97
20     192.168.10.128   255.255.255.224 /27     192.168.10.129-158  192.168.10.129

Hostname    Type                        VLAN    IP
DC2012      Server WIN2012VM (NIC-1)    10  192.168.10.7
FS2012      Server WIN2012VM (NIC-3)    10  192.168.10.8
VMM2012     Server WIN2012 (NIC-2)      10  192.168.10.6
VMM2012     Server WIN2012 (NIC-4)      15  192.168.10.102
WIN10PC1    Client WIN10                20  192.168.10.134
WIN10PC2    Client WIN10                20  192.168.10.135
WIN10PC3    Client WIN10                20  192.168.10.136
WIN10PC4    Client WIN10                20  192.168.10.137
RHEL6-1     Client RHEL6                20  192.168.10.142
RHEL6-2     Client RHEL6                20  192.168.10.143
RHEL6SRV    Server RHEL6 (NIC-1)        10  192.168.10.9
RHEL6SRV    Server RHEL6 (NIC-2)        15  192.168.10.103
RHEL6-3     Client RHEL6 (NIC-1)        10  192.168.10.10
RHEL6-3     Client RHEL6 (NIC-2)        15  192.168.10.104

On both RHEL6-3 and RHEL6SRV, I used the network manager and set the following route on the second NIC: 192.168.10.96 MASK 255.255.255.224 GATEWAY 192.168.19.97

On VMM2012, I set: route -p ADD 192.168.10.96 MASK 255.255.255.224 192.168.19.97 IF 17 (17 being NIC-4)

DC2012 and FS2012 are VM's physically located on VMM2012. They each use a dedicated NIC. In windows, I have the "Allow management operating system to share this network adapter" option unchecked.

All computer authenticate via DC2012.

All computers have CIFS mounts to FS2012 and VMM2012.

RHEL6-3 has NFS4 mounts to RHEL6SRV and CIFS mounts to VMM2012 via VLAN15

RHEL6SRV and VMM2012 have CIFS mounts to eachother via VLAN15

Problem 1: I receive the following messages in my log:

Jun 14 09:07:27 RHEL6-3 kernel: martian source 192.168.10.127 from 192.168.10.102, on dev eml
Jun 14 09:07:29 RHEL6-3 kernel: martian source 192.168.10.10 from 192.168.10.7, on dev em2
Jun 14 09:08:04 RHEL6-3 kernel: martian source 192.168.10.31 from 192.168.10.7, on dev em2
Jun 14 09:07:32 RHEL6-3 kernel: martian source 255.255.255.255 from 192.168.10.9, on dev em2
Jun 14 10:21:39 RHEL6SRV kernel: martian source 192.168.10.127 from 192.168.10.102, on dev eml
Jun 14 11:50:55 RHEL6SRV kernel: martian source 192.168.10.103 from 192.168.10.102, on dev eml
Jun 14 11:51:03 RHEL6SRV kernel: martian source 255.255.255.255 from 192.168.10.9, on dev em2
Jun 14 11:51:36 RHEL6SRV kernel: martian source 192.168.10.9 from 192.168.10.6, on dev em2

I tried (tcpdump -ni em2) on RHEL6SRV while viewing messages and found corresponding mesages:

ARP, Request who-has 192.168.10.9 tell 102.168.10.6, length 46
martian source 192.168.10.9 from 192.168.10.6, on dev em2

IP 192.168.10.9.43202 > 255.255.255.255.hlserver: UDP, length 40
martian source 255.255.255.255 from 192.168.10.9, on dev em2

I need to diagnose the source of my mis-configuration in my network

Problem 2:

When attempting to mount NFS4 or CIFS shares by IP address of the VLAN15 interface, I receive an "access denied by server while mounting" error but if I try to mount using the hostname (which maps to the VLAN10 IP), it works. I'm unsure how to handle DNS for the secondary IPs since the system has only one host name.

Example of Client /etc/fstab:

//192.168.10.102:/share  /winshare   cifs    noauto,sec=krb5i,multiuser,cifsacl,nosuid,nodev,noexec 0 0
192.168.10.103:/share  /nfsshare   nfs4    noauto,sec=krb5p,nosuid,nodev,noexec 0 0

Example of server NFS export:

/share    *(rw)
  • If martian packets are seen, it seems unlikely there are multiple VLANs/networks, and are putting several logical networks on top of a broadcast domain. What does tcpdump show when sniffing one or the other interface? – Rui F Ribeiro Jun 14 at 16:51
  • All the systems are connected to the same managed switch. The switch is configured to multiple VLANs. Each system's interface is setup with the given gateway settings. I don't know what is meant by "several logical networks on top of a broadcast domain". What would I look for in a tcpdump? – eng3 Jun 14 at 16:55
  • Well, if there are martian packets, it might mean someone forgot probably to configure the VLANs on the switching side (or they are configured as a trunk), and both in the martian logs, and in tcpdump, you well see several networks being shown in the same interface, instead of seeing only one. Is it possible to add interface configuration on the linux and switching side, and tcpdumps to the question? Otherwise we might be making just conjectures here. – Rui F Ribeiro Jun 14 at 17:00
  • Is there a specific configuration setting of the VLANs that should be reviewed? I can ask the network person who setup the switch. I assumed the the VLANs were setup if the gateways worked. I assumed that if it wasnt setup, I wouldnt have connectivity at all because the NICs would be pointing to non-existent gateways. – eng3 Jun 14 at 17:03
  • I advised adding the tcpdumps for starters. – Rui F Ribeiro Jun 14 at 17:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.