Dear customers,
== TL;DR
- Our colo provider is moving out of Telehouse to another datacentre close by,
and we are going to be moving with them.
- It will mean one short (2h or less) outage for each of your VMs as our
servers are physically moved on multiple dates (to be decided) in
December and January.
- Nothing about your service will change. IPs, specs, prices will all remain
the same. Our colo provider will retain the same transit providers and will
still peer at LINX.
== Background
Since we started in 2006 BitFolk has been hosted with the same colo provider in
the same datacentre: Telehouse London. Telehouse have decided to not renew our
colo provider's contract with them and so they will be moving almost all of
their infrastructure to another datacentre; the nearby IP House.
https://www.ip-house.co.uk/
Given that we have much less than a single rack of infrastructure ourselves,
our options here are to either move with our current colo provider or find
another colo provider. Staying exactly where we are is not an available option.
In light of the good relationship we have had with our colo provider since
2006, we have decided to move with them. This must take place before the middle
of January 2024.
== Planning
At this early stage only broad decisions have been made. The main reason I'm
writing at this time is to give you as much notice as possible of the
relatively short outage you will experience in December or January. As soon as
more detailed plans are made I will communicate them to you.
As soon as the infrastructure is available at IP House we plan to move some of
our servers there - ones with no customer services on them. We will schedule
the physical movements of our servers that have customers VMs on them across
multiple dates in December and January. As IP House is only about 500 metres
from Telehouse, we don't expect an outage of more than about 2 hours for each
server that is moved.
If you can not tolerate an outage like that, please open a support ticket by
emailing support(a)bitfolk.com as soon as possible. We will do our best to
schedule a live migration of your service from hardware in Telehouse to
hardware in IP House, resulting in only a few seconds of unreachability. You
can contact us about that now, or you can wait until the exact date of your
move is communicated to you, but there are a limited number of customers we can
do this for. So please only ask for this if it's really important to you. If it
is, please ask for it as soon as you can to avoid disappointment.
== Answers To Anticipated Questions
=== Will anything about my service change?
No. It will be on the same hardware, with the same IP addresses and
same specification for the same price.
We're aware that we're well overdue for a hardware refresh and we hope to be
tackling that as soon as the move is done. That will result in a higher
specification for the same price.
=== Will the network connectivity change?
No. Our colo provider will retain the same mix of transit providers that it
currently does, and it will still peer at LINX.
=== When exactly will outages affect me?
We have not yet planned exactly when we will move our servers. As soon as we do
we'll contact all affected customers. We need to co-ordinate work with our colo
provider who are also busy planning the movement of all of their infrastructure
and other customers, and installing new infrastructure at IP House.
We expect there to be several dates with one or more servers moving on each
date. All customers on those servers wil experience a short outage, but in
total we would expect only the one outage per VM.
=== Will I be able to change the date/time of my outage?
No. As the whole server that your VM is on will be moving at the specified
time, your options will be to either go with that or seek to have your service
live migrated ahead of that date. Please contact support if you want one or
more of your VMs to be live migrated.
If you have multiple VMs on different servers it is possible that they will be
affected at the same time, i.e. if the two servers that your VMs are on are
both to be relocated in the same maintenance window. If that is undesirable,
again one or all of your VMs will need to be live migrated to servers already
present in IP House.
=== What is live migration?
It's where we snapshot your running VM and its storage, ship the data across to
a new host and resume execution again. It typically results in only a few
seconds of unreachability, and probably TCP connections still stay alive.
More information: https://tools.bitfolk.com/wiki/Suspend_and_restore
== Thanks
Thanks for your custom. Despite the upheaval I am looking forward to
a new chapter in a new datacentre!
Andy Smith
Director
BitFolk Ltd
--
https://bitfolk.com/ -- No-nonsense VPS hosting
Does anyone else use Docker on their VPS and if so what spec is it running on? I'm migrating to a new one to upgrade to 64bit and decided to try Docker. It has worked pretty well on my local servers, none massively powerful, but in spite of being told the overhead of multiple copies of things isn't much of an impact it seems I have obliterated the capability of my VPS barely half way through the process, with htop registering both swap and physical memory maxed out.
I've migrated 8 WordPress sites and had to disable 3 to regain some level of stability (these using the official WordPress Docker image, which admittedly spins up Apache and MariaDB for each one). I've decommissioned 2 WordPress sites, but have 7 more to move (4 of those being in 2 multisite setups) as well as 3 phpBB, 3 Piwigo and a Roundcube, not to mention email and mailman.
These are low traffic and experimental sites in many cases, but at the moment I'm debating between reverting to the old setup (possibly an issue as my phpBB sites are planned to migrate to Discourse which started the Docker idea), or moving the databases into a single container, and possibly the WordPress sites into a single Apache (which also part defeats the object of Docker I guess).
I was expecting more load, but not this much. Part way through is not an ideal time, particularly when I'm still battling COVID two and a half weeks on from catching it (which isn't helping my thought processes!).
Any thoughts from the collective wisdom here?
Paul Tansom | Aptanet Ltd. | https://www.aptanet.com/ | 023 9238 0001
=============================================================================
Registered in England | Company No: 4905028 | Registered Office: Ralls House,
Parklands Business Park, Forrest Road, Denmead, Waterlooville, Hants, PO7 6XP
Hi,
I'm currently investigating problems with host "macallan" which
started having issues around 12:57 as far as I can see. I will keep
you updated when I know more,
Thanks,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
I've just installed a new VPS using the ubuntu-jammy image, and with the
new (likely not that new, but last VPS was pretty old) cloud-init
install it creates a default user of ubuntu. I want to change that to
something else, but haven't needed to do that before, so thought I'd
check if there are any gotchas that will come back to bite me down the
line. As far as I can tell the best way to do it is to use:
usermod -l newusernam -m -d /home/newusername oldusername
Clearly from a different account, and with this being a clean new
install there shouldn't be anything in terms of cron files to mess with.
There will be a group name to change too though. Is there anything else,
or a better way to do this?
--
Paul Tansom | Aptanet Ltd. | https://www.aptanet.com/ | 023 9238 0001
=============================================================================
Registered in England | Company No: 4905028 | Registered Office: Ralls House,
Parklands Business Park, Forrest Road, Denmead, Waterlooville, Hants, PO7 6XP
Hello all,
Would any of you know if the following scenario is "doable"?
We run an old Exchange 2010 infrastructure at my work, and there is no way
they are going to spring for newer: getting them to go from 2003 to 2010
was an ordeal...
Could I set up an Ubuntu Postfix "relay" server between Exchange and the
Internet, that also permits one particular mailbox to be accessible from a
Dovecot install on the same server (as well as relaying the mail for that
mailbox to Exchange)?
Yes/no and pointers most welcomed.
Kind regards
Murray Crane
