Packet loss between two specific VMs

hadooplocal-area-networknetworkingpacketlossping

I am experiencing some high-percentage packet loss (more than 70%) between two specific VMs in a local network, in a remote cluster (I have no physical access to it).

One machine is used as a hadoop master and the other as a hadoop slave.
The strange thing is that when I ping any other machine from/to any of these two machines, I get no packet loss. Network configurations for all the slave nodes are identical.

Best Answer

It seems that the problem is fixed. Here are the two things I did:

I changed the etc/network/interfaces file (in Ubuntu) and added my local network with a static IP for my machine, like this:

auto eth1
iface eth1 inet static
address 192.168.0.18
netmask 255.255.255.0
gateway 192.168.0.1`

The new IP of my VM (192.168.0.18) is different than the previous one (192.168.0.2).

Any comments on how did that solve my problem?

Related Solutions

Linux – Packet loss between Broadcom NetXreme NIC’s on different hosts

Here's Serverfaults official answer on the matter: http://blog.serverfault.com/post/broadcom-die-mutha/

Ping a Specific Port

You can't ping ports, as Ping is using ICMP which is an internet layer protocol that doesn't have ports. Ports belong to the transport layer protocols like TCP and UDP.

However, you could use nmap to see whether ports are open or not

nmap -p 80 example.com

Edit: As flokra mentioned, nmap is more than just a ping-for-ports-thingy. It's the security auditers and hackers best friend and comes with tons of cool options. Check the doc for all possible flags.

Best Answer

Related Solutions

Linux – Packet loss between Broadcom NetXreme NIC’s on different hosts

Ping a Specific Port

Related Topic