The timeline and details are as follows:

* GeekISP’s hosting services will terminate officially on January 31, 2021.
* Final archival backups will be taken starting approximately February 1, 2021 and stored for 3 months. These archives are intended for extraordinary circumstances.
* Machines will be powered down once the archival backup is complete. This will likely proceed in phases.
* Registrar reseller services via Dotster will continue indefinitely. Your domain ownership is safe!
* Customers who have paid for service beyond January 31, 2021 will receive a prorated refund by PayPal or check.

In terms of transitioning your account to a new hosting company, there are many capable and reasonably priced options. Here is a list of possible options that I considered for my own usage:

Hosting:

* wordpress.com
* render.com
* Amazon Web Services (caveat: requires a nontrivial effort)

Mail Services:

* PolarisMail
* ProtonMail

I will unfortunately not be able to perform a migration of your site or email. I would recommend reaching out to whatever company you choose for assistance transitioning your information and of course, I will be available to answer any questions.

Thank you for your business over the years. It has been a pleasure working to provide hosting services for you!

During the work inbound mail to accounts at GeekISP will be delayed. Webmail, desktop, and mobile mail clients will likely report errors. Follow us on Twitter (@GeekISP) for updates along the way.

EDIT: Maintenance is complete as of approximately 9:50pm ET. Disks were swapped and we’re on a new zpool. Please email support@geekisp.com if you have any issues or questions.

During the maintenance window, all mail-related services will be offline. Follow us on Twitter (@GeekISP) for updates along the way.

It’s not clear that this maintenance will directly fix the issues that we’ve seen lately. The hope is that if we’re able to boost the IO performance of the primary mail volume by reconfiguring it slightly, then we’ll relieve some of the pressure in the system and things can return to normal. Absent the underlying cause, it appears to be our best and easiest option at the moment.

Update: 12:01am EST – Mail service has been restored. The original plan had to be scrapped as it was simply taking too long to clone the mail volume to the secondary server. Instead an upgrade to FreeBSD 10 was performed and this went smoothly. Several additional improvements were also planned but not yet executed in an effort to minimize the total outage. These include migrating from courier imap to dovecot as well as moving authentication data into MySQL.

We will likely need to perform another service interruption at some point in order to execute the original plan of changing our disk layout, but the aforementioned newly planned items can likely be handled, at least mostly, while still online.

1. Authentication errors (“Invalid password”, “Password rejected”, or similar)
2. Timeouts or other very slow operations retrieving mail, deleting mail, or similar

Unfortunately this far the exact underlying cause of the issue has not surfaced. As this has gone on for some time and appears to be happening with increasing frequency and severity, I’m exploring alternative arrangements that would address the issue, even without a full understanding of the problem.

Please accept my apologies for the inconvenience this has caused.

1. VPS servers – most will be offline.
2. SMTP mail services – all customers will not be able to send or receive mail during the maintenance window. You can still check mail and compose drafts.
3. SVN services – all repositories will be offline.

I expect the maintenance will require approximately 20-30 minutes to complete. We’ll be replacing a disk backplane and adding additional memory.

Sorry for any inconvenience this causes, and as before, follow @GeekISP on twitter for updates.

We’re currently scheduled to move at roughly 6am US/Eastern time on Saturday July 20th. We expect all services to be offline during the move for approximately 4 hours. More updates and more estimates will be posted as we get closer. For updates during the migration, follow us on Twitter.

Since that discovery I’ve been trying to find effective mitigation techniques, but have had to resort to periodic blocks of all requests to that URL. My apologies for any inconvenience this may have caused. As of yet I don’t have a complete resolution to the issue, but will post updates when possible.

This move will be a “lights out” cutover. E.g. we’re going to power down the existing equipment, disconnect it, toss it in a truck, and drive it to the new facility. Unfortunately that means we’ll be offline for the duration of the cutover, but it is the only realistic way to do it given GeekISP’s current architecture. We’re expecting about 2-3 hours of downtime during the move, and this will affect all services at GeekISP. For email, don’t worry – your mail will be queued on the sending side and delivered once we’re back online. We’re looking into the possibility of having our provider send HTTP traffic to a maintenance page during the move and if possible, we’ll do that as well.

We’re currently targeting May 5th at 7am to start the move, but this is not completely final. Updates will be posted if the date or time changes.

As always – please contact support@geekisp.com if you have any questions.

UPDATE 1 – May 1st: Word from Gotham is that there are some issues with routing between the current and the new datacenter. We will not be migrating on May 5th, and are waiting for updates before rescheduling.