We are looking to get a new VPS.

The VPS comes with 1 IPv4 as standard. We have always previously used metal box servers containing BIND nameserver / DNS functionality within. These have had 2 distinct IPs (sometimes regrettably on the same subnet mask) .

I have searched the internet and it is overall very ambiguous and far too general for giving me useful pointers. However, this Question is close to what I'm asking: How to setup my own nameservers with a shared DNS service on my VPS

However, that is from 2012 and also doesn't fully answer my query;

I believe from using DNS on metal boxes that I need two distinct IPs for the nameserver ns1.server.com and ns2.server.com and I can't see any clear indicator if this remains true for VPS servers.

• Is my belief correct, eg we need to have two distinct IPs for best practise for the same VPS machine ?

Bonus:

• Is there any benefit in having the second/additional nameserver IP as an IPv6 rather than as an IPv4 [in UK]?

There are many similar questions on here but nothing that looks like it's actually what I need.

I know what I need is very simple but I can't websearch it, probably I'm missing a key word.

My issue:

There is a standard DNS set up on a third party (as yet unseen by my eyes) controlling the website and email. Defined as "Server Old". There is a new website set up on its own server, defined "Server New" this server hosts the website ONLY and so I need to update the DNS of the original "server Old" DNS hoster.

I understand I should update the original A record (only) to the IP of the server hosting the new website.

But I am having trouble understanding what to do about the DNS for the new-website server, it needs to recognise when dealing with email to send emils out to the WAN rather than dealing with them internally. How do I set up this DNS, for example if the new website sends emails to the website owner, the email needs to be send from the new server to the old server;

Is it simply a case of updating the MX record on the new-website server to point back to the original DNS server?

Or have I missed some things here?

Update 1

In response to Tilman's answer to my above query. A kinda rewrite and clarification response to the above answer.

There is a Domain; "Domain Alpha" that has it's A, MX and all other records set correctly.

1. I believe that, in order to redirect the website -only- to a new server, that on the DNS rules set for "Domain Alpha" I should update the "A" DNS rule -only- to the IP of the new website server.

2. I would then have a DNS ruleset on this new server that would -only- handle incoming HTTP requests and anything else be pushed back to the original FQDN DNS ruleset.

3. Basically, I want to delegate all services EXCEPT email (MX) to the new server and leave the email as it is.

Can I do this (3) by setting an A ruleset record in the existing DNS ruleset that points to the new server, and then on the new server (which also contains DNS rulesets) setting an explicit MX record back to the "original" current server IP (or the orginal / current DNS host)?

I am looking for help with a Linux server (CentOS) guide or script that can be used to send an email to a server account when a new SFTP connection is detected. For example, giving the connection name and the requesting IP address as well as connection authentication type (if possible) (such as SSH Key or a Password, etc.) .

I have next to no experience with Bash scripts, however I have successfully made exactly this which detects SSH connections; however I can't find out where to go to extend this to also notify the email address of SFTP Connections to the server,

Many thanks for any help on this.

I do realise this is pretty crappy question and I apologise, but search engine results are giving me lots of false results such as "how do I SFTP to my server?!" etc. which are clearly inappropriate.

Cheers

Update 1

So the /var/logs/secure collects the SFTP connection information. I would like to be able to grab that information some how and throw it out in a basic email.

My current working SSH detector does this in .bashrc:

echo -e 'ALERT - SSH access detected:' `date` `ls -l \`tty\` | awk '{print $3}'` '\n\nConnection Details: ' `w -h` '\n\nList of WHO: ' `who --login` | mail -s "Alert: Server Access Email Subject" email@address.com

This is called,triggered by CSF (ConfigServerFirewall)

I would like somehow to combine the two above so that when a info line is added to the /var/log/secure such as with:

Subsystem       sftp    /usr/libexec/openssh/sftp-server -l INFO

Then it will also be able to detect the line starts:

Accepted publickey for ....

That this line can then be thrown in an email out to the email address....

Update 2:

I may need to write my own Bash shim ....

Update 3:

Thanks to Piotr, my code for my shim is now:

#!/bin/bash
# Create a temporary log file
LOGFILE=$(/bin/mktemp /tmp/sftplog.XXXXXX)
# Redirect stderr to LOGFILE
exec 2>"$LOGFILE"

# Run the SFTP with logging to stderr
/usr/libexec/openssh/sftp-server -e -u 022 -l VERBOSE

# Use some sendmail substitute to send an e-mail
/usr/sbin/sendmail -i root@localhost <<EOF
From: email@localaddress.co.uk
To: email@localaddress.co.uk
Subject: SFTP connection for user $(LOGNAME)

Hello,
User $(LOGNAME) just connected to the SFTP server from $(SSH_CONNECTION).

Connection log:
$(<"$LOGFILE")
EOF

# echo -e "Hello,\nUser $(LOGNAME) just connected to the SFTP server from $(SSH_CONNECTION).\n\nConnection log:\n$(<"$LOGFILE")" | mail -s "SFTP connection for user $(LOGNAME)" email@localaddress.co.uk

# Delete the log
rm -f "$LOGFILE"

I have run the sendmail instruction from the command line and this works correctly, however new SFTP connections are resulting in EOF while reading packet.

Update 4

Reducing the script to :

#!/bin/bash
# Create a temporary log file

# Run the SFTP with logging to stderr
/usr/libexec/openssh/sftp-server -e -u 022 -l INFO
exec >/dev/null

Still returns the EOF while reading packet issue when connecting.

Update 5:

Setting the file permissions to be identical to the permissions of the original subsystem file (/usr/libexec/openssh/sftp-server) resolves the issue and the script runs correctly.

Very simple question that I can't find a reliable answer to, with Google -- can my server be the reverse DNS for multiple IP addresses?

Is it bad practise / wrong / asking for trouble to say my server sub.hostname.co.uk is the rDNS for:

118.168.10.100

and

118.168.11.200

NOTE: These IP addresses are in different subnets (.10. and .11. )

This Q&A seems to imply it would cause awkwardness. To clarify further, the second IP is a secondary IP, if that changes the requirement criteria.

Each Site on the server will be associated primarily with one of the two IPs.

?? Thanks!

I am trying to clarify my Reverse DNS PTR lookup for a domain. I have a server with a single IP addresses and multiple accounts, all sharing that IP address;

My PTR is currently:

 24.210.31.in-addr.arpa. 14400 IN PTR domainname.co.uk.

But the domain that the IP address reverses to is ns1.domainnetwork.co.uk, therefore should the PTR record on the account reference the primary returned domain rather than the account domain?

I can't seem to find any clear guidance on this with Google searching, however this question did seem to suggest that the account should reference the server domain rather than the account specific domain.

So, would having :

24.210.31.in-addr.arpa. 14400 IN PTR ns1.domainnetwork.co.uk.

in the DNS for the account give it a valid PTR record, even though this doesn't mention the account domain at all?

Edit: To clarify, the ns1.domainnetwork.co.uk is the primary name server for the domain.

UPDATE + EDIT

I have a persistent issue with this topic, and felt it best to resurrect this incompleted query than to start a new one from scratch.

All the above still holds true.

I have a server with various domains, on an IP address .xx.xx.xx.236 and xx.xx.xx.238 . I also have a single domain on its own IP address just for that doman (for the TLS certificate), which is xx.xx.xx.241 but I can't get that domain to correctly show a vaid PTR record in the DNS.

A domain on the server that works (ip address 12.34.56.236 ):

websitename.co.uk       14400 IN  A  12.34.55.236
55.34.12.in-addr.arpa.  14400 IN PTR nameserver1.network.co.uk.

the above works and gives a valid PTR feedback.

The nameserver1.network.co.uk. is the same primary name server as the domain which sits on its own IP address (12.34.56.241) but which never gives a correct PTR record.

The domain that does not give out a valid PTR record (unique IP address):

websitenameTwo.co.uk    14400 IN  A  12.34.55.241
55.34.12.in-addr.arpa.  14400 IN PTR nameserver1.network.co.uk.

What Am I missing to make this record have a valid PTR DNS record?