I recently joined a Facebook group, “WordPress Hosting”, to gain a better understanding of the hosting market, what my competitors are doing.
What I’ve learned is there is one large company buying up smaller hosting companies and generally providing very poor service after they acquire them but they are paying up to $200/customer to acquire these companies.
In response to this a lot of small companies have sprung up offering re-seller accounts and they will pay re-sellers for new accounts up to close to that price with the intent of once they acquire a large enough customer base, selling out.
They make money, the customer with the long term contract for what they thought would be quality hosting gets screwed.
I’ve also learned the majority of these providers use cpanel, and the majority of them don’t execute individual sites code with their own UID so when one site gets hacked the whole lot of them do. Problems with defaced sites and Trojan software run rampant on these sites.
Because of these sites WordPress has gotten a bad rap.
This is not my model. Those of you that have been with us know that Eskimo North has remained independently owned since ten years before the Internet even existed. I’m not here just trying to collect numbers to sell out. I’m here to offer you reliable secure consistently high performing service over the long haul.
If you have someone design a website for you, find out if they’re recommending a hosting company based upon a commission they are getting or based upon the performance you will get. If they’re recommending someone else, odds are good it’s the former.
I awoke from a dream this morning, and that’s why I am up now, the dream was too disturbing emotionally to not get up for a while at least.
It was my last day of high school. It was so entirely real. The emotional reality of knowing I would never see many of these people again was horrible and that emotion lingers. Because my wife works at night, I’ve got nobody to share it with right now, except you folks.
I just got a phone call at 5:45 AM, where the phone rang and the caller ID said, “StarNet Inc”, and the number was 941-347-6089.
StarNet Inc was an old 56k wholesale dial up provider I used before they folded and were acquired by Global POPs. That WAS there number but it’s been disconnected for years, and calling it just gets re-order.
If you are using Frontier for your Internet Access, and having problems sending e-mail using our servers, I discovered an issue today and a work-around.
Frontier’s DNS servers are returning the IP address of a proxy server instead of our real e-mail address when you do a DNS query for mail.eskimo.com. The proxy server proxies web and incoming mail servers but does not pass or proxy the outgoing ports.
The work around for this is to go into your network connection configuration and select manually specify name servers and specify name servers that are not Frontiers, such as ours, 126.96.36.199 and 188.8.131.52 or open DNS or Google’s or some other public DNS servers.
I have filed a bug report on Launchpad so hopefully this will get addressed. This is a bug specific to Ubuntu or at least Debian releases, was not a problem with Redhat.
According to the manual page, restarting libvirtd is not supposed to interrupt running virtual machines, but it does. Thus when an upgrade to libvirtd comes out on the host machines, if I apply it, it frequently restarts the guest machines which is unpopular with customers to put it mildly.
This is how to reproduce it, create kvm-qemu guests, with a dozen of them or so running on a box, on the host machine type systemctl restart libvirt-bin and on average about 3/4ths of the guest machines will reboot.
Sorry for the unintended reboot around 11:45AM today. There was an update to libvirtd and every time Ubuntu updates it, it restarts it. Restarting it is NOT supposed to interrupt service per the manual:
Restarting libvirtd does not impact running guests. Guests continue to
operate and will be picked up automatically if their XML configuration
has been defined. Any guests whose XML configuration has not been
defined will be lost from the configuration.
But unfortunately, for about half the guest machines it does, regardless of what the manual states.
I will be rebooting the host computers and some of the guests to load new kernels tonight at approximately 10PM. Hopefully it will go smoothly like last time and result in only minimal downtime (about 10 minutes per machine).
Also, one of the two radius servers is currently down and will be restored at that time. This should not impact service as radius is duplicated and the machine that is down is the secondary, just some loss of redundancy until it is restored.
I rebooted the web server to load a new kernel and it did not automatically mount the owncloud data directory which is NFS mounted from a file server.
Occasional failure to mount an NFS mount point is a bug that currently seems to exist in all of the debian based distributions. Not having the same problem with the Redhat based distributions. Not sure if it is a systemd issue or kernel issue or just what yet.
I’ve just manually mounted it, should be good now.