Cannot ping other servers

[hjazz6]

Hi,

I have an IBM Bladecenter HS21 with 3 blade servers. All 3 blade servers were initially installed with RedHat Linux 4. I then installed ixMOS on one of the blade servers, and Mandriva Linux was also installed (over RedHat) automatically along with ixMOS.

Now, I have some network problems. The other 2 servers with RedHat can ping each other, but it appears that the server with ixMOS/Mandriva Linux is isolated, that is, I can't ping the other 2 servers from it, and I can't ping it from the other 2 servers.

Does anyone have any ideas on what is wrong and how I can solve this problem? One more thing, my installation of Mandriva Linux has no GUI, so I have to do everything using the command line interface.

Thank you.

[sarlacii]

Hi hjazz6... we need a little more info to go on. LOL Firstly, are your ifaces up and running (netstat -i -ev)? Then, what routing do you have in place (i.e. did you ifaces get their proper ips etc. from DHCP - or were they manually specified)? And also, try a traceroute to see where the ping is going... may be routing out into the wild, then trying to come back (if you have multiple gateways). Need info.

[hjazz6]

I have eth0 and lo up on all three servers.

The IP addresses for all 3 servers are manually configured:

Mandriva server: 192.168.168.202
RedHat #1: 192.168.168.201
RedHat #2: 192.168.168.203

I'm not at the servers now, so I'll try traceroute later.

[hjazz6]

My Mandriva server is not installed with traceroute. But I tried traceroute with the 2 RH servers and here's what I got:

From RedHat #1 to #2:

traceroute to 192.168.168.203 (192.168.168.203) 30 hops max, 46 byte packets
1   192.168.168.203 (192.168.168.203)  0.114ms  1.362ms  0.098ms

From RedHat #1 to Mandriva:

traceroute to 192.168.168.202 (192.168.168.202) 30 hops max, 46 byte packets
1   192.168.168.202 (192.168.168.202)  3000.487ms !H  3000.496ms !H  2999.988ms !H

From RedHat #2 to #1:

traceroute to 192.168.168.201 (192.168.168.201) 30 hops max, 46 byte packets
1   192.168.168.201 (192.168.168.201)  0.114ms  1.362ms  0.098ms

From RedHat #2 to Mandriva:

traceroute to 192.168.168.202 (192.168.168.202) 30 hops max, 46 byte packets
1   192.168.168.202 (192.168.168.202)  3000.775ms !H  3000.715ms !H  3000.978ms !H

One more thing I found out.

I looked at /etc/sysconfig/network-scripts/ifcfg-eth0 of the Mandriva server, which contains the following:

DEVICE = eth0
BOOTPROTO=static
IPADDR=192.168.168.202
NETMASK=255.255.255.0
GATEWAY=192.168.168.1
ONBOOT=yes
METRIC=5
MII_NOT_SUPPORTED=yes
USERCTL=no
DNS1=192.168.168.2
RESOLV_MODS=yes
LINK_DETECTION_DELAY=6
IPV6INIT=no
IPV6to4INIT=no

and noticed that if the line MII_NOT_SUPPORTED equals "yes", then the server has the IP address 192.168.168.202 upon startup, but if MII_NOT_SUPPORTED=no, then the server has no IP address upon startup, and I have to manually assign one.

Also, when I typed "mii-tool" on the Mandriva server (regardless of MII_NOT_SUPPORTED=yes or =no), I get the following:

SIOCGMIIPHY on 'eth0' failed: Operation not supported
SIOCGMIIPHY on 'eth1' failed: Operation not supported
SIOCGMIIPHY on 'eth2' failed: Resource temporarily unavailable
SIOCGMIIPHY on 'eth3' failed: Resource temporarily unavailable
no MII interfaces found

But on the 2 RedHat servers, which didn't contain the MII_NOT_SUPPORTED lines, I get the following output:

eth0: negotiated 10baseT-HD, link ok
SIOCGMIIPHY on 'eth1' failed: Resource temporarily unavailable
SIOCGMIIPHY on 'eth2' failed: Operation not supported
SIOCGMIIPHY on 'eth3' failed: Operation not supported

I ran the ethtool program and got the following output.

output of "ethtool -t etho" on Mandriva server

The test result is FAIL
The test extra info:
nvram test (online)     0
link test (online)         1
register test (offline)   0
memory test (offline)   0
loopback test (offline)  0
interrupt test (offline)  0

output of "ethtool -t etho" on both RedHat servers

The test result is PASS
The test extra info:
register test (offline)   0
memory test (offline)   0
loopback test (offline) 0
nvram test (online)     0
interrupt test (online)  0
link test (online)         0

output of "ethtool eth0" on Mandriva server:

Settings for eth0:
   Supported ports: [ FIBRE ]
   Supported link modes: 1000baseT/Half 1000baseT/Full
   Supports auto-negotiation: Yes
   Advertised link modes: 1000baseT/Half 1000baseT/Full
   Advertised auto-negotiation: Yes
   Speed: Unknown! (65535)
   Duplex: Unknown! (255)
   Port: FIBRE
   PHYAD: 1
   Transceiver: internal
   Auto-negotiation: on
   Supports Wake-on: g
   Wake-on: d
   Current message level: 0x000000ff (255)
   Link detected: no

output of "ethtool eth0" on RedHat servers:

Settings for eth0:
   Supported ports: [ FIBRE ]
   Supported link modes: 1000baseT/Full
   Supports auto-negotiation: Yes
   Advertised link modes: 1000baseT/Full
   Advertised auto-negotiation: Yes
   Speed: 1000 Mb/s
   Duplex: Full
   Port: FIBRE
   PHYAD: 2
   Transceiver: internal
   Auto-negotiation: on
   Supports Wake-on: g
   Wake-on: g
   Link detected: yes

Does this mean that there are some problems with the eth0 interface of the Mandriva server?

Thank you.

[sarlacii]

Hi hjazz6... traceroute shows that you are not routing your queries out of the network into a blackhole... the trace effectively just times out on the first leg, so routing seems fine, but your iface is down. Okay, so on to the next test result... the mii-tool has been replaced by ethtool, so that is why the line "MII_NOT_SUPPORTED" is missing from Redhat server configs and is set to "yes" on the Mandriva box.

However, ethtool has told you that you have a problem... and you can guess what it is from the results. Firstly, the rather obvious "no link". Secondly, and more interestingly, assuming that the hardware on all three servers is identical, you should surely get the same output from all three servers when you run "ethtool eth0"?!? The fact that the Mandriva box reports different results means that either the driver is wrong on that box (since ethtool simply interrogates the iface via the driver) or the iface is bust.

Run "dmesg" and check for your iface... I get:

skge 1.6 addr 0xd8020000 irq 20 chip Yukon-Lite rev 7
skge eth0: addr 00:50:70:45:11:6d
skge eth0: enabling interface
ADDRCONF(NETDEV_UP): eth0: link is not ready
skge eth0: Link is up at 100 Mbps, full duplex, flow control tx and rx
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

So, my device driver is recognised as "skge"... I then check /proc/bus/pci and /proc/modules (use lspci and lsmod commands to get human readable output on those two files, i.e. automatically looks at device and vendor ids etc.) to make sure that the correct drivers are loaded for the iface.

lspci: 02:01.0 Ethernet controller: Marvell Technology Group Ltd. 88E8001 Gigabit Ethernet Controller (rev 13), and lsmod: skge 38128 0

Now, because you have two working servers, you can cheat and check the Redhat boxes to see what the correct driver should be... what is the Mandriva system using compared to the Redhat systems?

Other than that... your hardware could be broken such that the mnandriva system doesn't detect it correctly, and assigns the wrong driver?

Good luck mate.

[hjazz6]

Turns out the Mandriva server did load the wrong driver.

I have 4 ethernet interfaces eth0 - eth3. On the RedHat servers, eth0 and eth1 use the BCM5708 driver, while eth2 and eth3 use BCM95704. On the Mandriva server, eth0 and eth1 use BCM95704 and eth2 and eth3 use BCM5708. After I switched the drivers to the correct interfaces, all 3 servers can now ping each other using eth0. However, eth2 and eth3 cannot be used to ping each other on all 3 servers. So I guess the servers can't support BCM95704?