userpal's questions

I have a NAS which has 2 10Gbe ports, and it provides Adaptive Load Balancing (ALB) link aggregation function.

By checking Wikipedia, it seems that in ALB, the bonding driver needs to go through complicated processes (intercept ARP, overwrite HW addresses, etc) to make the link aggregation works.

I plan to connect both network ports to a 10Gbe switch and use the ALB function to increase the bandwidth, but i worry that the ALB function will cause high CPU usage.

Will ALB link aggregation causes high CPU usage in a NAS? Will it really able to increase the bandwidth to >10Gbps without affecting the NAS performance?

nohup <command> <arg> &

When I SSH into a Linux server, if I want to run a command and make sure that it will continue to run in the background after I logout from SSH, I will use the above commands.

Recently I am using a server stack called Bitnami Node.js stack. It is a self-contained software bundle. There is a node binary in the bin folder in the installation directory.

In the command line, I am able to use the node command to run my JS programs: node <js_program.js>. There is no problem.

But, I would like to continue to run the program after I logout from SSH. When I run nohup <installation_dir>/nodejs/bin/node <js_program.js> > <stdout.out> &, the program will run in the background. But, when I logout from SSH, the program will also terminate immediately. It seems that the nohup command has no effect in this case.

How can I fix this issue?

Let's say in a server A with 100GB partition, user create a NFS export directory. In server B, user mount the NFS share to use it.

In server A, can the NFS server limit the amount of disk space that the NFS client can use? Or the client will always be able to use the full disk space available?

I am new to Docker, but I have experience of using VM and hypervisor. Referring to Docker introduction and this image, Docker containers are isolated but share the same host OS.

If we deploy only 1 Docker container (e.g. LAMP app) in a host OS, can the Docker container uses the full CPU power and memory of the host OS?

Is there any performance difference between a LAMP app running in a Docker container and a LAMP app running in the host OS directly?

I am studying a framework called Apache Mesos. I am totally new to this framework.

According to this website, it can combine your datacenter servers and cloud instances into one shared pool from which any app or framework can draw, as illustrated by this image.

I have read some articles and tutorials about this framework. But I am still not very clear how this framework works.

Let's say I have 3 physical server machines with 4 cores and 8GB RAM each. Can I deploy a single LAMP web application and let it uses the combined hardware resources of all servers (12 cores and 24GB RAM)?

By referring to this tutorial, it seems that Mesos can only run the web application in one of the servers. Then, what are the benefits of using Mesos?

There is a Linux server with 3 network interfaces, eth0, eth1, eth2. IP forwarding has been turn on in this server.

1. eth0 is connected to 10.0.1.0/24. Its IP is 10.0.1.1.
2. eth1 is connected to 172.16.1.0/24. Its IP is 172.16.1.1. Server A can ping router C at 172.16.1.2.
3. eth2 is connected to 192.168.1.0/24. Its IP is 192.168.1.1. Server A can ping server B at 192.168.1.2.
4. Router C is able to route to 172.16.2.0/24 and 172.16.3.0/24.

                                          [10.0.1.0/24]
                                                |
172.16.2.0/24------|                            |
                   [C]------172.16.1.0/24------[A]------192.168.1.0/24------[B]
172.16.3.0/24------|

We have set MASQUERADE at eth0. When server B (192.168.1.2) connect to 10.0.1.0/24, IP MASQUERADE will happen at eth0.

Can we set MASQUERADE at eth1? Is it okay to set MASQUERADE at more than 1 network interfaces in Linux?

I have installed openssh-server in my Ubuntu PC using apt-get install openssh-server. The version is 5.9.

Now, I would like to compile and install openssh-server version 6.2 from source codes. I have successfully downloaded the source codes, and run the following commands:

./configure make make install

I found that the new version of openssh-server was installed into /usr/local/sbin/. The old version of openssh-server is in /usr/sbin/.

I found that the service script in /etc/init.d/ssh is still pointing to /usr/sbin/. And the old openssh-server (v5.9) is still running.

1. How can I replace the old openssh-server with the new openssh-server that I have just compiled and installed?
2. How can I create a init.d script to start and stop the new openssh-server that I've compiled from source manually?
3. How to start the new openssh-server on boot?
4. When I install openssh-server using apt-get install, the config files will be installed into /etc/ssh/. If I compile and install it from source, where is the config file?
5. If I compiled openssh-server from source, but I install openssh-client package using apt-get install, will there be any config files conflict?

Thanks.

I am new in using virtualization technologies. I have read about Puppet and Foreman. If I am not mistaken, these software are used to automate the process of installing and configuring servers, for both real machines and virtual machines. An example is shown in this video.

Refer to this website, in a cloud computing environment, we can create VM template and use it to deploy many VM with similar configurations. A VM template is a master copy of a virtual machine that can be used to create and provision virtual machines.

Refer to the Youtube video mentioned above, if we use Puppet and Foreman to install and configure VM, it seems that we need to repeat the installation and configuration process in all VM. Why can't we just use VM template?

Hi I am new to Fedora/Centos/RHEL. I am confused about NetworkManager and network service in these Linux OS. May I know what is the relation between NetworkManager and network service?

If I am using NetworkManager, do i need to turn off network service? Similarly, if I am using network service, do I need to turn off NetworkManager? Will they conflict with each other if I use both application at the same time?

I am new to Apache web server. When we configure Apache web server, we have the options of setting name-based virtual host.

Based on my understanding, when a user type in a domain name www.example.com in a web browser, it will connect to a DNS server which will convert the domain name to a (public) IP address. Then, the web browser will connect to the web server using the IP address.

Now, let's say I have a server with a public IP address, a.a.a.a. I have registered 2 domain names, www.example.com and www.example1.com. Both of these 2 domain names are pointing to the same IP address, a.a.a.a. So, when user types in www.example.com or www.example1.com in his web browser, the DNS server will point them to the IP address a.a.a.a.

I have configured name-based virtual host according to the user guide at this web site.

However, what I don't understand is: Since the DNS server already converted domain names to IP address, all users will connect to the web server using IP address a.a.a.a. How Apache knows whether the user is using www.example.com or www.example1.com in their web browser?

I am new to Windows networking. When people talk about TCP tuning on Windows platform, they always mention about TCP Window Size. I am wondering whether Windows uses the concept of "Socket Buffer Size"?

On Windows XP, the TCP window size is fixed. We can set it using the TCPWindowSize registry value. How about Socket Buffer Size? How can we set Socket Buffer size on Windows? Can we set it to a value different from TCP window size?

I want to use Winpcap to capture all network packets going through a Gigabit NIC of a server.

Assuming that I am able to utilize the network link up to 100%, the maximum network speed is 1000Mbps. If we exclude the TCP/IP headers, the maximum TCP data rate should be roughly 940Mbps.

Let's say I send a 1GB file through the NIC at 940Mbps using TCP destination port 6000. I use Winpcap to capture all network packets going through the NIC and then dump it to a pcap file. If I use Wireshark to analyze the pcap file and then check the sum of packet size for all network packets sent to TCP port 6000, am I able to get exactly 1GB from the pcap file?

Thanks.