JYelton's questions

I run an instance of MySQL server (5.6), port 3306, on an Ubuntu (14.04 LTS) VM. Employees use a client application to interact with the database.

Periodically, I get a troubleshooting call where the user cannot connect, with the error:

Unable to connect to any of the specified MySQL hosts.

The MySQL error log does not contain a refused or denied connection for the user.

Restarting the machine (not just MySQL) resolves the problem.

I've determined that this appears to be due to client machines being assigned a different IP address. I've also determined that restarting the Ubuntu networking service also resolves the problem.

I am suspicious that there is a network security feature at work here, but I am unable to figure out what it is. I would like to know:

• Can I disable this "feature" or at least configure it to trust clients on the LAN?
• Is there a way to view client machines' IP addresses that are being blocked by this mechanism?

Edit:

Firewall setup: ufw is inactive, iptables I believe is "factory default"

I'd like to join a client computer to a domain via the internet. Ideally I just want a basic VPN connection so that the client can see the domain controller. The DC is running Windows Server 2012 Essentials, but apparently the only way to create a VPN is by using Anywhere Access.

Anywhere Access wants me to use a domain name and an SSL certificate, which I don't want to do. This server is not hosting a web site nor is there a domain that I want to use. I just want to connect using the public WAN address long enough to get the client joined to the domain. This is a temporary thing, not a remote work situation.

Is there some other way to set up a VPN on Windows Server 2012 Essentials? The client computer is running Windows 7 Pro.

I have a problem in which I cannot access a server using its public name from the same network as the server. Access to the server works fine from the internet.

Reading up on this problem, I've ran across such things as hairpin NAT, loopback NAT, split DNS, editing hosts files, etc.

My network has a SMC 8013WG-CCR (Comcast) cable modem connected to a Linksys WRT54G2. As I understand, the router is supposed to handle this sort of loopback (by leaving the security option "Filter Internet NAT Redirection" disabled), but the connection between these two devices is a secondary LAN, so I don't think the router "knows" what the proper public IP address is.

The easiest solution would be to edit the hosts files of all the computers in the network, but many are notebook computers which will need to access the server both on the LAN and externally.

The server is Windows Server 2012, so I could set it up as the internal DNS server but I don't have enough experience with this to really understand how it would interact with the router, DHCP, etc.

I am prepared to purchase a new router if I need something with greater capability. It appears from my research that most IT admins prefer to solve this problem with split DNS rather than hairpin NAT. I understand the concept of hairpin NAT more clearly than split DNS. It seems to me that the router just needs to provide LAN clients with the local IP of the server when clients use the public name (sub.domain.com). Isn't that just an entry in a local DNS cache? Is that what split DNS does?

What would be a good solution to implement this fix for a small business?

Short version:

What software can I use to mirror data (files/folders) between two Windows servers across a broadband internet connection?

Longer version:

I'd like to implement an off-site backup solution for a company with two locations. Instead of transporting external drives to/from the locations or employee homes, I am interested in implementing a data mirroring solution that will use two servers to back each other up. The idea is that in the event of a physical catastrophe, one server should have the data for both locations.

Both servers will be running Windows Server 2008 or newer and both will have disks arranged in a RAID for local redundancy and drive failure protection.

I had encountered a software program at one time that was designed to accomplish this mirroring by limiting the bandwidth used on an internet connection during business hours, and doing the majority of data mirroring at night. The idea being that data is copied to the other server as soon as it is changed, but in such a way that bandwidth usage does not compromise business activities. Unfortunately I cannot remember nor find such a software program and am looking for advice from the ServerFault community.

I am using WampServer 2.0i (PHP 5.3.0 and Apache 2.2.11) on Windows* and would like to connect to SQL Server 2008. I downloaded the SQL Server Driver for PHP 1.1 from Microsoft which provides an assortment of driver DLL's.

The chart I am provided with in the help files is:

Driver file                  PHP version    Thread safe?    Use with PHP .dll
===========                  ===========    ============    =================
php_sqlsrv_53_nts_vc6.dll    5.3            no              php5.dll
php_sqlsrv_53_nts_vc9.dll    5.3            no              php5.dll
php_sqlsrv_53_ts_vc6.dll     5.3            yes             php5ts.dll
php_sqlsrv_53_ts_vc9.dll     5.3            yes             php5ts.dll

Further instructions say:

If the name of the driver file contains "vc9", it should be used with a PHP version compiled with Visual C++ 9.0. If the name of the driver file contains "vc6", it should be used with a PHP version compiled with Visual C++ 6.0.

How am I supposed to know which version of Visual C++ was used to compile PHP? I realize on a Unix distro I'd have more intimate knowledge of this because I'd likely be installing PHP and Apache separately and compiling them (it's been ages since I've done so).

Should I even try to install this driver dll, or is there some better way to get PHP to talk to SQL Server?

*Windows 7 Professional 64-bit; SQL Server 2008; WampServer 2.0i

I am running Wampserver 2.0i on Windows 7 64-bit.

I'm not experiencing any problems with any of the processes, but I noticed that mysqld is maintaining over 32,000 open handles. Is this normal behavior for this process?

I am also running MS SQL Server, and its process (sqlservr) only has 400+ handles.