Chau Chee Yang's questions

Case 1: SSH host has ONE network interface

Routing table for host SSH:

[SSH] $ ip route
default via 192.168.211.1 dev eth0
192.168.211.0/24 dev eth0 proto kernel scope link src 192.168.211.119

Initiate ssh connection from SSH_Client to host SSH on LAN_0 works fluently:

[SSH_CLIENT] $ ssh [email protected] -p 22

Case 2: SSH host has TWO network interface

Routing table for host SSH:

[SSH] $ ip route
default via 192.168.0.1 dev eth1
192.168.0.0/24 dev eth1 proto kernel scope link src 192.168.0.113
192.168.211.0/24 dev eth0 proto kernel scope link src 192.168.211.119

Initiate ssh connection from SSH_Client to host SSH on LAN_0 works fine:

[SSH_Client] $ ssh [email protected]

Initiate ssh connection from SSH_Client to host SSH on LAN_211 works for few seconds and freeze thereafter:

[SSH_Client] $ ssh [email protected]

I am looking for a solution to let a private LAN HTTP service accessible from WAN with minimal configuration.

For example, I have more than 1000 deployments of HTTP service across nation running in LAN. These HTTP service are not available to public at the moment.

Now, the HTTP services would like to serve public request. Configure port forwarding for these HTTP services is tedious.

One near possible solution I can think of is setup a traffic server (or reverse proxy server) in WAN. These 1000 HTTP services will find a way to register the traffic server and stay connected all the time via some kind of reverse connection of tunneling . The traffic server will get request from public and route the request to corresponding HTTP service in backend and response the request to public.

After some research, I learned the reverse proxy Apache Traffic Server or Amazon CloudFront or Azure's CDN service may serve the front end well.

However, the backend HTTP services that could only achieve via some kind of reverse connection or tunneling doesn't support well by the front end traffic server.

Please share some opinions that may helps for this case. Thank you.

I have a number of ready to use network appliances like router, print server, L2/L3 network switches, ip-pbx, access point and many others.

These devices are usually managed by web portal by accessing port 80 (http). And asking for user name and password for accessing.

With so many devices to manage, I usually use default user name and password without changing it in case I might forgot in future. However, this habit is risky and easy for someone to change the configuration easily.

Is there a way to leave the default username / password as is but prevent other local network users to do the changes on the configuration?

Can VLAN segmentation help by grouping all those devices in one VLAN? But doing so seems prevent workstations from other VLAN accessing the services of those devices.

Or may be firewall + VLAN may help?

Cat 6A cable is thicker than Cat6 cable. For a Cat6 cable and connector, we can plug into any network switch that has RJ45 connector even a 100Mbps network switch.

Just wonder if I use Cat6A cable and Cat6A connector, can I plug it into the normal RJ45 network port? Will it fit since it's cable size if much thicker.

I have 30 workstation and 30 IP Phones all under one single network 192.168.9.0. The network switches is unmanaged switch but supports IEEE 802.1p QoS (e.g.: DLink DGS 1008D and DGS 1016D).

I learned that it is better to split the data and voice network to avoid lagging of voice traffic by using VLAN and managed switch.

However, if the network switch support QoS, do I still need to configure VLAN to split the traffic?

When I use Samba SWAT to change the smb.conf setting, I can't find include statement.

I then try manually change smb.conf:

[test]
  path = /mnt/test
  available = no
  include = /etc/samba/smb.%G.conf

Assume the group is "users", the /etc/samba/smb.users.conf contain:

  available = yes

The samba share "test" works as expected. I then login to SWAT but I can't find include statement setting. Making changes in SWAT will remove my include statement.

I am using samba 3.5.8-75 in Fedora Core 13.

I use autofs to mount some iso files in folder /mnt/iso:

Fedora-13-x86_64-DVD.iso
Fedora-14-x86_64-DVD.iso
Fedora-15-x86_64-DVD.iso

The autofs is working fine and I can see the contents of iso files:

# ls /mnt/iso/fedora.13/
# ls /mnt/iso/fedora.14/
# ls /mnt/iso/fedora.15/

After finish the autofs testing. I force unmount all iso folders:

# umount -l /mnt/iso/*

I then export the /mnt/iso in /etc/exports:

/mnt/iso        192.168.0.0/24(ro,insecure,crossmnt,all_squash)

The NFS service is working fine too. I may mount the /mnt/iso from another machine via nfs client:

# mount -t nfs4 <nfs-server>:/mnt/iso /mnt

I can see all iso mount in /mnt/ folder:

# ls /mnt
dr-xr-xr-x 15 nobody nobody 4096 2011-06-11 10:23 .
drwxr-xr-x 24 root   root   4096 2011-02-08 08:45 ..
drwxr-sr-x  7 nobody nobody 4096 2010-05-13 09:06 fedora.13
drwxr-sr-x  7 nobody nobody 4096 2010-10-22 02:32 fedora.14
drwxrwsr-x  7 nobody nobody 4096 2011-05-14 03:50 fedora.15

However, when I try to list the contents for each folder:

#  ls /mnt/fedora.13

Nothing shown on screen. Using command like this

#  ls /mnt/fedora.13/*

prompt

ls: cannot access /mnt/fedora.13/*: No such file or directory

The only case I success to list out the content of the iso folders in nfs client machine is using this steps:

Unmount nfs share from client machine:

# umount /mnt

using ls to show all sub-trees of /mnt/iso in nfs server:

# ls /mnt/iso/*

The autofs will mount all iso files automatically.

Mount the nfs folder in client machine:

# mount -t nfs4 <nfs-server>:/mnt/iso /mnt

List the sub-trees of /mnt:

# ls /mnt/*

You may then see all the contents in nfs client machine.

However, this defeat the purpose of autofs, as I need to show the contents once in the autofs machine first.

Does anyone has any ideas? I am using the following:

1. Fedora 13 x86_64
2. autofs-5.0.5-28
3. nfs-utils-1.2.2.2-2

I am using Hylafax 6.0.5 on Fedora 13 x86_64. As there are no rpm package available for Fedora 13, I use the source tar ball to install hylafax myself. Everything seems fine during compile and install.

I try to send a fax with sendfax and encounter error:

# sendfax -n -d <fax-number> /etc/passwd
/usr/local/sbin/textfmt: No font metric information found for "Courier-Bold".
Usage: /usr/local/sbin/textfmt [-1] [-2] [-B] [-c] [-D] [-f fontname] [-F fontdir(s)] [-m N] [-o #] [-p #] [-r] [-U] [-Ml=#,r=#,t=#,b=#] [-V #] files... >out.ps
Default options: -f Courier -1 -p 11bp -o 0
Error converting document; command was "/usr/local/sbin/textfmt -B -f Courier-Bold      -Ml=0.4in -p 11 -s default >'/tmp//sndfaxp5GdJ9' <'/etc/passwd'"

It seems like there is problem with font problem. I have ghostscript-fonts installed too.

I can't find hyla.conf in path /etc/hylafax. There is no /etc/hylafax path in my file system. All configuration files seems located in /var/spool/hylafax/etc.