15 Dec
2016
15 Dec
'16
3:20 p.m.
Hi,
As you may be aware there are a few things that we (or third party
services on our behalf) scan for on our own network. A
non-exhaustive list of these things are:
- Open SNMP servers
- Open DNS resolvers
- Open portmapper services
- Open MongoDB / Memcached / Elasticsearch / Redis
- Open Remote Desktop
- Open multicast DNS (Avahi)
- Open TFTP server
- SSLv3/Poodle vulnerable services
…and so on.
Where these can only negatively affect the operator of the service
(e.g. SSLv3/Poodle or an open MongoDB), we are content to just email
you forever.
However, many of these problems are worth scanning for because they
are easily and frequently used to attack other hosts. For example,
any "chatty" UDP protocol like portmapper or DNS can receive spoofed
requests which will amplify, making for a DDoS on a third party.
So, when we email customers about these it's because we need them
fixed.
Unfortunately some customers are either not receiving these emails
or are happy to ignore them, for months at a time, basically until
we open a support ticket with them and ask why they aren't dealing
with it. This is taking up too much human time.
"We did not receive the email" is now no longer a valid excuse
because there is provision in our web panel for an
emergency/alternate contact:
https://panel.bitfolk.com/account/contacts/#toc-address-book
If we've been emailing a customer for weeks about this then we will
have also sent a copy to the emergency/alternate contact at some
point.
So, I would like your opinions on how you think we should deal with
this. Two proposals I can think of are:
a) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, a
firewall rule will be added to block the problematic service and
an invoice will be raised for a managed firewall service, which
will be a monthly recurring charge. This will be quite expensive.
Or:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I have already broached this question on IRC and basically no one
was in favour of (a) because they did not feel that surprise
invoices would go down well.
If you have other suggestions for how it should be handled I would
be happy to read and consider them. Ideas that involve either not
dealing with the problematic services or that aren't suitable for
automation are not likely to be acceptable though.
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
15 Dec
15 Dec
3:25 p.m.
On Thu, Dec 15, 2016 at 03:20:08PM +0000, Andy Smith wrote:
Hi,
As you may be aware there are a few things that we (or third party
services on our behalf) scan for on our own network. A
non-exhaustive list of these things are:
- Open SNMP servers
- Open DNS resolvers
- Open portmapper services
- Open MongoDB / Memcached / Elasticsearch / Redis
- Open Remote Desktop
- Open multicast DNS (Avahi)
- Open TFTP server
- SSLv3/Poodle vulnerable services
…and so on.
Where these can only negatively affect the operator of the service
(e.g. SSLv3/Poodle or an open MongoDB), we are content to just email
you forever.
However, many of these problems are worth scanning for because they
are easily and frequently used to attack other hosts. For example,
any "chatty" UDP protocol like portmapper or DNS can receive spoofed
requests which will amplify, making for a DDoS on a third party.
So, when we email customers about these it's because we need them
fixed.
Unfortunately some customers are either not receiving these emails
or are happy to ignore them, for months at a time, basically until
we open a support ticket with them and ask why they aren't dealing
with it. This is taking up too much human time.
"We did not receive the email" is now no longer a valid excuse
because there is provision in our web panel for an
emergency/alternate contact:
https://panel.bitfolk.com/account/contacts/#toc-address-book
If we've been emailing a customer for weeks about this then we will
have also sent a copy to the emergency/alternate contact at some
point.
So, I would like your opinions on how you think we should deal with
this. Two proposals I can think of are:
a) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, a
firewall rule will be added to block the problematic service and
an invoice will be raised for a managed firewall service, which
will be a monthly recurring charge. This will be quite expensive.
Or:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I have already broached this question on IRC and basically no one
was in favour of (a) because they did not feel that surprise
invoices would go down well.
If you have other suggestions for how it should be handled I would
be happy to read and consider them. Ideas that involve either not
dealing with the problematic services or that aren't suitable for
automation are not likely to be acceptable though.
I'm fairly happy with (b) however it'd be nice to be more specific about
what can get you suspended, maybe visibility in the panel of your scan
status, that sort of thing.
Michael
3:45 p.m.
Hi Michael,
On Thu, Dec 15, 2016 at 03:25:02PM +0000, Michael Stevens wrote:
I'm fairly happy with (b) however it'd be nice
to be more specific about
what can get you suspended,
At the moment the scanning is done either by me running a script or
else by ShadowServer running something every day and emailing us a
CSV. Based on those, we then send out emails asking customers to fix
it.
Right now, Portmapper is the most common one and there is a
semi-automated "please fix this" email that goes out. I will also
send out "please fix this" emails for SNMP and DNS resolver because
they are also DDoS amplification vectors. But they are less
automated as they are rarer.
So anyway what I am saying is: there's a small number of things,
which we would of course list off on some wiki page and automate the
(email) reporting of.
maybe visibility in the panel of your scan status,
that sort of
thing.
I hadn't, however, intended going as far as making Nagios alerts and
bits on the panel etc for this, as that would be significant work. I
was really hoping that email alerts would be enough for this. I'm
also willing to SMS people who've supplied a cell phone number the
day before the network suspension will happen.
Is that sufficient?
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
5:18 p.m.
On Thu, 2016-12-15 at 15:45 +0000, Andy Smith wrote:
I'm not sure putting the scan status on the panel would be worth much -
surely as soon as anything comes up you'll be receiving an email
anyway?
maybe visibility in the panel of your scan
status, that sort of
thing.
I hadn't, however, intended going as far as making Nagios alerts and
bits on the panel etc for this, as that would be significant work. I
was really hoping that email alerts would be enough for this. I'm
also willing to SMS people who've supplied a cell phone number the
day before the network suspension will happen.
Is that sufficient?
Cheers,
Andy
3:44 p.m.
Is it possible to do a (without charging) for 21 days in the hope of
getting their attention and then b?
On 15 Dec 2016 3:20 pm, "Andy Smith" <andy(a)bitfolk.com> wrote:
Hi,
As you may be aware there are a few things that we (or third party
services on our behalf) scan for on our own network. A
non-exhaustive list of these things are:
- Open SNMP servers
- Open DNS resolvers
- Open portmapper services
- Open MongoDB / Memcached / Elasticsearch / Redis
- Open Remote Desktop
- Open multicast DNS (Avahi)
- Open TFTP server
- SSLv3/Poodle vulnerable services
…and so on.
Where these can only negatively affect the operator of the service
(e.g. SSLv3/Poodle or an open MongoDB), we are content to just email
you forever.
However, many of these problems are worth scanning for because they
are easily and frequently used to attack other hosts. For example,
any "chatty" UDP protocol like portmapper or DNS can receive spoofed
requests which will amplify, making for a DDoS on a third party.
So, when we email customers about these it's because we need them
fixed.
Unfortunately some customers are either not receiving these emails
or are happy to ignore them, for months at a time, basically until
we open a support ticket with them and ask why they aren't dealing
with it. This is taking up too much human time.
"We did not receive the email" is now no longer a valid excuse
because there is provision in our web panel for an
emergency/alternate contact:
https://panel.bitfolk.com/account/contacts/#toc-address-book
If we've been emailing a customer for weeks about this then we will
have also sent a copy to the emergency/alternate contact at some
point.
So, I would like your opinions on how you think we should deal with
this. Two proposals I can think of are:
a) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, a
firewall rule will be added to block the problematic service and
an invoice will be raised for a managed firewall service, which
will be a monthly recurring charge. This will be quite expensive.
Or:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I have already broached this question on IRC and basically no one
was in favour of (a) because they did not feel that surprise
invoices would go down well.
If you have other suggestions for how it should be handled I would
be happy to read and consider them. Ideas that involve either not
dealing with the problematic services or that aren't suitable for
automation are not likely to be acceptable though.
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
iEYEAREDAAYFAlhStKgACgkQIJm2TL8VSQu/gACfYuicZ4XdWkRloHOk2iiZqvoZ
4zAAoPNkiMEPcWN/dVVxDPKzUDtpfqBk
=9rkS
-----END PGP SIGNATURE-----
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
3:50 p.m.
Hi Robert,
On Thu, Dec 15, 2016 at 03:44:04PM +0000, Robert Gauld wrote:
Is it possible to do a (without charging) for 21 days
in the hope of
getting their attention and then b?
I don't think that would get anyone's attention as all of these
services tend to be used from the VPS itself and aren't actually
intended to be publicly available. A firewall rule applied on
BitFolk's host would not impact normal use in most cases.
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
4:35 p.m.
Andy Smith <andy(a)bitfolk.com> said, in message
20161215152008.GJ1804(a)bitfolk.com:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
Option b) doesn't sound excessive to me. When I was looking after
student networking at a Uni, we suspended network access for
misbehaving hosts with *much* less effort than that! Might be
worth your while publicising the fact that people can register
an emergency contact though. I'd not noticed this or registered
one myself (now fixed).
Option a) would (unless I'm missing it) require a change of your
terms and conditions and presumably a non-responsive contact could
claim they were never told about the changed T&C's. So, unless you
actually enjoy the task of chasing someone for cash they don't
want to pay, option a) probably isn't going to make your life
any easier!
Cheers,
Alun.
--
Dr. Alun Jones, auj(a)aber.ac.uk, 01970 622637
Mathemateg, Ffiseg a Chyfrifiadureg, Prifysgol Aberystwyth
Mathematics, Physics & Computer Science, Aberystwyth University
9:12 p.m.
On 15/12/2016 17:20, Andy Smith wrote:
a) After at least 21 days of sending email alerts to
the main
contact and the emergency contact and receiving no response, a
firewall rule will be added to block the problematic service and
an invoice will be raised for a managed firewall service, which
will be a monthly recurring charge. This will be quite expensive.
Or:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I personally prefer (a) with sms notification but I think overall (b)
would be more effective for you. Charging people who don't monitor
contact emails would probably mean they won't see the invoice either.
Then, when you eventually manually hunt them down or their contract term
ends, they'll have a mountain of debt and will probably dispute it.
I would expect that it must be stressed more prominently to the
customers that it's their responsibility to keep their information
up-to-date and what will happen if they don't. Maybe a periodical stock
email reminder "is your information up-to-date? go here to check". You
could piggy-back some kind of mini-newsletter on that, too. E.g. the xen
fixes.
--GM
10:12 p.m.
I would vote for option 2) as well. It's pretty common that if you breach terms of
service, your service gets discontinued, temporarily or permanently. For example on AWS if
your instances are found to be part of a DDoS attack, the instances are immediately ring
fenced on the network until the owner can be spoken to and corrective action can be taken.
I also think that suspending all networking on the VPS, i.e. the service they are paying
you for, will be the most effective way of getting their attention.
Kind regards,
Paul
Sent from my iPad. Please excuse brevity, spelling and punctuation.
On 15 Dec 2016, at 21:13, G. Miliotis
<corfiot@elementality.org<mailto:corfiot@elementality.org>> wrote:
On 15/12/2016 17:20, Andy Smith wrote:
a) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, a
firewall rule will be added to block the problematic service and
an invoice will be raised for a managed firewall service, which
will be a monthly recurring charge. This will be quite expensive.
Or:
b) After at least 21 days of sending email alerts to the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I personally prefer (a) with sms notification but I think overall (b) would be more
effective for you. Charging people who don't monitor contact emails would probably
mean they won't see the invoice either. Then, when you eventually manually hunt them
down or their contract term ends, they'll have a mountain of debt and will probably
dispute it.
I would expect that it must be stressed more prominently to the customers that it's
their responsibility to keep their information up-to-date and what will happen if they
don't. Maybe a periodical stock email reminder "is your information up-to-date?
go here to check". You could piggy-back some kind of mini-newsletter on that, too.
E.g. the xen fixes.
--GM
_______________________________________________
users mailing list
users@lists.bitfolk.com<mailto:users@lists.bitfolk.com>
https://lists.bitfolk.com/mailman/listinfo/users
19 Dec
19 Dec
4:57 a.m.
On Thu, Dec 15, 2016 at 10:20 AM, Andy Smith <andy(a)bitfolk.com> wrote:
Unfortunately some customers are either not receiving
these emails
or are happy to ignore them, for months at a time, basically until
we open a support ticket with them and ask why they aren't dealing
with it. This is taking up too much human time.
"We did not receive the email" is now no longer a valid excuse
because there is provision in our web panel for an
emergency/alternate contact:
https://panel.bitfolk.com/account/contacts/#toc-address-book
If we've been emailing a customer for weeks about this then we will
have also sent a copy to the emergency/alternate contact at some
point.
Well, now I have some first-hand experience with this. :(
On Fri Dec 16 Andy filed a RT ticket with me about spam being sent from my
account. Unfortunately, he quoted the problematic spam in the RT ticket,
which means that his email to me included the spam verbatim, which of
course meant that it ended up in my spam folder as well and I never saw it.
On Sun Dec 18 (two days later) there was another report of spam, Andy
notified me via RT that he was going to turn off networking. Of course RT
included context, which included the entire problematic spam email again,
so of course I didn't see it either.
He then turned off networking for my domain. Which certainly got my
attention.
As it turns out, this was a simple misconfiguration on my part -- I had
whitelisted relaying from a specific IP range back in 2011, but after the
project ended I didn't maintain control of that IP block... and apparently
it has since falled into the wrong hands. I could have fixed this quickly
with prompt notice.
But that's neither here nor there. What my story seems (to me) to indicate
is that "we did not receive the email" is in fact a valid excuse if the
email in question quotes spam. I didn't have alternate contacts listed,
but I could have listed a dozen and they would most likely all have
classified Andy's email (10% "what is this" and 90% quoted spam) as spam.
I had followed this conversation on the users list, but assumed it would
not be relevant to me, since I was receiving all of Andy's emails here on
the list without problem. Content-specific filters make that a bad
assumption.
I think there is need for a non-email backup. Sending an SMS would have
been guaranteed to get through, I bet.
--scott
5:02 a.m.
On Sun, Dec 18, 2016 at 11:57 PM, C. Scott Ananian <cscott(a)cscott.net>
wrote:
I think there is need for a non-email backup. Sending
an SMS would have
been guaranteed to get through, I bet.
Mea culpa: there is in fact already the ability to add a phone number as an
emergency contact. I've now done so, and would encourage everyone else to
do so as well.
--scott
--
( http://cscott.net/ )
9:36 a.m.
On Mon, Dec 19, 2016 at 12:02:01AM -0500, C. Scott Ananian wrote:
On Sun, Dec 18, 2016 at 11:57 PM, C. Scott Ananian
<cscott(a)cscott.net>
wrote:
Where is this capability? I don't see it.
I think there is need for a non-email backup.
Sending an SMS would have
been guaranteed to get through, I bet.
Mea culpa: there is in fact already the ability to add a phone number as an
emergency contact. I've now done so, and would encourage everyone else to
do so as well.
12:28 p.m.
Hi Adam,
On Mon, Dec 19, 2016 at 09:36:11AM +0000, Adam Spiers wrote:
On Mon, Dec 19, 2016 at 12:02:01AM -0500, C. Scott
Ananian wrote:
When you add a contact record, one of the fields is for phone
number.
https://panel.bitfolk.com/account/contacts/#toc-address-book
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
Mea culpa: there is in fact already the ability to
add a phone
number as an emergency contact. I've now done so, and would
encourage everyone else to do so as well.
Where is this capability? I don't see it.
12:37 p.m.
On Mon, Dec 19, 2016 at 12:28:02PM +0000, Andy Smith wrote:
Hi Adam,
On Mon, Dec 19, 2016 at 09:36:11AM +0000, Adam Spiers wrote:
Nope, I already checked there, but definitely no phone number field.
I already tried adding a new contact too, via
https://panel.bitfolk.com/account/contacts/add
and again, no phone number field. This is the same with both accounts
I have access to.
On Mon, Dec 19, 2016 at 12:02:01AM -0500, C.
Scott Ananian wrote:
When you add a contact record, one of the fields is for phone
number.
https://panel.bitfolk.com/account/contacts/#toc-address-book Mea culpa: there is in fact already the ability to
add a phone
number as an emergency contact. I've now done so, and would
encourage everyone else to do so as well.
Where is this capability? I don't see it. 
12:42 p.m.
On Mon, 19 Dec 2016, at 12:37, Adam Spiers wrote:
Nope, I already checked there, but definitely no phone
number field.
I already tried adding a new contact too, via
https://panel.bitfolk.com/account/contacts/add
and again, no phone number field. This is the same with both accounts
I have access to.
The pull down has extra fields that you can add. Default (showing) is
company name, but phone numbers are in the list.
Robert
--
Robert McWilliam rmcw(a)allmail.net www.ormiret.com
Apparently, three out of four people make up 75% of the population.
12:45 p.m.
Hello,
On Mon, Dec 19, 2016 at 12:42:51PM +0000, Robert McWilliam wrote:
On Mon, 19 Dec 2016, at 12:37, Adam Spiers wrote:
That's obviously not clear enough. How could it be improved?
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
https://panel.bitfolk.com/account/contacts/add
and again, no phone number field. This is the same with both accounts
I have access to.
The pull down has extra fields that you can add. Default (showing) is
company name, but phone numbers are in the list.
12:49 p.m.
On 19 December 2016 at 12:45, Andy Smith <andy(a)bitfolk.com> wrote:
Hello,
On Mon, Dec 19, 2016 at 12:42:51PM +0000, Robert McWilliam wrote:
Just ditch the pull-down? It's not as if the add contact page is too
long and needs to be kept short.
There's another problem: there doesn't seem to be any way to add phone
numbers to an
existing contact, only to a new one. Surely I don't have to delete a
contact and re-add it,
just to add a phone number?
On Mon, 19 Dec 2016, at 12:37, Adam Spiers
wrote:
That's obviously not clear enough. How could it be improved?
https://panel.bitfolk.com/account/contacts/add
and again, no phone number field. This is the same with both accounts
I have access to.
The pull down has extra fields that you can add. Default (showing) is
company name, but phone numbers are in the list.
12:55 p.m.
Hello,
On Mon, Dec 19, 2016 at 12:49:11PM +0000, Adam Spiers wrote:
On 19 December 2016 at 12:45, Andy Smith
<andy(a)bitfolk.com> wrote:
How would you then pick amongst the four (and soon to be more)
different things you could add?
What about if the initial text was "Select detail to add…" instead of
it starting off with the suggestion of "Company name…" ? Thus
inviting you to click.
That's obviously not clear enough. How could
it be improved?
Just ditch the pull-down? It's not as if the add contact page is too
long and needs to be kept short. There's another problem: there doesn't seem to
be any way to add phone
numbers to an
existing contact, only to a new one. Surely I don't have to delete a
contact and re-add it,
just to add a phone number?
Underneath the contact is a "change" button. Does that not let you
add a phone number? It does for me.
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
12:59 p.m.
On 19/12/16 13:55, Andy Smith wrote:
Works for me as well.
I also agree with Richard about a dedicated field for a mobile-number
with suggestions of SMS-notification.
- OM
Hello,
On Mon, Dec 19, 2016 at 12:49:11PM +0000, Adam Spiers wrote:
Yes, that would help.
On 19 December 2016 at 12:45, Andy Smith
<andy(a)bitfolk.com> wrote:
How would you then pick amongst the four (and soon to be more)
different things you could add?
What about if the initial text was "Select detail to add…" instead of
it starting off with the suggestion of "Company name…" ? Thus
inviting you to click. That's obviously not clear enough. How could
it be improved?
Just ditch the pull-down? It's not as if the add contact page is too
long and needs to be kept short. There's another problem: there doesn't
seem to be any way to add phone
numbers to an
existing contact, only to a new one. Surely I don't have to delete a
contact and re-add it,
just to add a phone number?
Underneath the contact is a "change" button. Does that not let you
add a phone number? It does for me.
1:20 p.m.
On Mon, Dec 19, 2016 at 12:55:24PM +0000, Andy Smith wrote:
Hello,
On Mon, Dec 19, 2016 at 12:49:11PM +0000, Adam Spiers wrote:
Just have all the fields immediately available.
On 19 December 2016 at 12:45, Andy Smith
<andy(a)bitfolk.com> wrote:
How would you then pick amongst the four (and soon to be more)
different things you could add? That's obviously not clear enough. How could
it be improved?
Just ditch the pull-down? It's not as if the add contact page is too
long and needs to be kept short. What about if the initial text was "Select detail
to add…" instead of
it starting off with the suggestion of "Company name…" ? Thus
inviting you to click.
Ah, I see that now thanks. But again that's pretty unintuitive to me.
I would expect that button to change the /association/ between a
contact and the alerting role, not to change the contact's details.
And also the most conspicuous place to change the details of the main
contact is in the first section, so it doesn't seem logical that you
can only change certain fields via that section, and not add phone
numbers etc.
It looks like the schema contains tables for the main contact,
secondary contacts, roles, and the role_contact association between
roles and secondary contacts. Is that right? If so I think that's
the source of some of the confusion, because all contacts should be
equal for the most intuitive UI, which would also allow the primary
and secondaries to be interchangeable.
I'd suggest asking a UI/UX designer how to do this better.
There's another problem: there doesn't
seem to be any way to add phone
numbers to an
existing contact, only to a new one. Surely I don't have to delete a
contact and re-add it,
just to add a phone number?
Underneath the contact is a "change" button. Does that not let you
add a phone number? It does for me.
1:49 p.m.
Hello,
On Mon, Dec 19, 2016 at 01:20:54PM +0000, Adam Spiers wrote:
Where would you expect the button to change a contact's details to
be if not directly below the contact, next to the button for
deleting it?
On Mon, Dec 19, 2016 at 12:55:24PM +0000, Andy Smith
wrote:
I don't like the idea but it seems to be what people want so I'm
prepared to accept I'm wrong.
How would you then pick amongst the four (and soon
to be more)
different things you could add?
Just have all the fields immediately available. Underneath the
contact is a "change" button. Does that not let you
add a phone number? It does for me.
Ah, I see that now thanks. But again that's pretty unintuitive to me.
I would expect that button to change the /association/ between a
contact and the alerting role, not to change the contact's details. And also the most conspicuous place to change the
details of the main
contact is in the first section, so it doesn't seem logical that you
can only change certain fields via that section, and not add phone
numbers etc.
I was hoping to eventually get rid of that main contact, turning it
into just another contact. So as a consequence I'm not keen to add
anything new to it (like phone numbers).
The main contact used to be all we had, so quite a lot of existing
code just gets your name and email address out of that. To save
having to rewrite everything at once I've left the main contact in
existence but if it is confusing then I will find a way to remove it
or at least not display it like it is.
It looks like the schema contains tables for the main
contact,
secondary contacts, roles, and the role_contact association between
roles and secondary contacts. Is that right? If so I think that's
the source of some of the confusion, because all contacts should be
equal for the most intuitive UI, which would also allow the primary
and secondaries to be interchangeable.
You're right in that the main contact existed first and so exists as
a separate thing. All the other contacts are "equal".
I'd suggest asking a UI/UX designer how to do this
better.
We can't really afford it so unfortunately will rely on free advice
from users.
It would have been good to get some of this input 5 months ago when
this feature was being worked on:
http://lists.bitfolk.com/lurker/message/20160723.011237.83603d29.en.html
but now is better than never, and 25 phone numbers have been added
just today, which is more than in the entire previous time to date,
so thanks for your email, Scott. :)
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
2:30 p.m.
On Mon, Dec 19, 2016 at 1:49 PM, Andy Smith <andy(a)bitfolk.com> wrote:
I was hoping to eventually get rid of that main
contact, turning it
into just another contact. So as a consequence I'm not keen to add
anything new to it (like phone numbers).
The main contact used to be all we had, so quite a lot of existing
code just gets your name and email address out of that. To save
having to rewrite everything at once I've left the main contact in
existence but if it is confusing then I will find a way to remove it
or at least not display it like it is.
It makes sense to me now that you've said that. I think it's confusing
as it is. For those people with only a single person on the account I
imagine there is a tendency to ignore the "add another contact" option
because I'm already the contact and so there should be no need.
Cheers,
Roger
2:32 p.m.
On Mon, Dec 19, 2016, at 02:30 PM, Roger Light wrote:
It makes sense to me now that you've said that. I
think it's confusing
as it is. For those people with only a single person on the account I
imagine there is a tendency to ignore the "add another contact" option
because I'm already the contact and so there should be no need.
This. At some point I probably clicked to add a new contact, saw that
there were no obvious fields that linked to anything different that
wasn't in the main contact, and clicked Cancel.
12:53 p.m.
On Mon, Dec 19, 2016, at 12:45 PM, Andy Smith wrote:
That's obviously not clear enough. How could it be
improved?
Given that there are only five options for fields to add, it seems
overly-complicated - just include them all on the form (although why
distinguish between home/work/other numbers?).
Personally I'd be tempted to include at least one phone field on the
main contact. Perhaps a specific field for a mobile number, making it
clear that it may be used for SMS notifications (as in incentive to
provide it)?
12:46 p.m.
On Mon, Dec 19, 2016 at 12:42:51PM +0000, Robert McWilliam wrote:
On Mon, 19 Dec 2016, at 12:37, Adam Spiers wrote:
Ah, that was quite well hidden ;-) Thanks!
Nope, I already checked there, but definitely no
phone number field.
I already tried adding a new contact too, via
https://panel.bitfolk.com/account/contacts/add
and again, no phone number field. This is the same with both accounts
I have access to.
The pull down has extra fields that you can add. Default (showing) is
company name, but phone numbers are in the list.
12:26 p.m.
Hi Scott,
On Sun, Dec 18, 2016 at 11:57:58PM -0500, C. Scott Ananian wrote:
On Fri Dec 16 Andy filed a RT ticket with me about
spam being sent from my
account. Unfortunately, he quoted the problematic spam in the RT ticket,
which means that his email to me included the spam verbatim, which of
course meant that it ended up in my spam folder as well and I never saw it.
Yes, apologies again for that rather rookie mistake. I really should
have realised that a message with spam in it might not get through.
He then turned off networking for my domain. Which
certainly got my
attention.
If the only issue were "getting the customer's attention" then I
would normally be happy to wait a lot longer - at least 21 days as
we've established now. In this case though I was seeing spam being
delivered so I had to shut off the network, and would have done that
whether we'd been in conversation or not.
But that's neither here nor there. What my story
seems (to me) to indicate
is that "we did not receive the email" is in fact a valid excuse if the
email in question quotes spam.
Me quoting spam was certainly a bad idea and I will avoid doing that
for at least the initial message in future.
When I say not receiving the email is not a valid excuse though,
it's specifically because I've now implemented multiple contact
methods one of which being phone. I will certainly make use of a
cell phone number to send a text message, before shutting anyone's
service off.
I will also add more contact methods if they seem popular and
relevant. Perhaps Twitter? I may not be able to direct message
people but can at least leave a "@yourname please contact us
urgently" message.
Content-specific filters make that a bad assumption.
It also highlights the fact that the concept of a "spam folder" can
be problematic is an outright delivery rejection would be more
useful to me than a delivery which may just go to a folder that's
never read. I know this is not always under people's control. But
the battle for email is lost and email deliverability is only going
to get worse.
Cheers,
Andy
--
https://bitfolk.com/ -- No-nonsense VPS hosting
1:17 p.m.
On 19/12/2016 06:57, C. Scott Ananian wrote:
But that's neither here nor there. What my story
seems (to me) to
indicate is that "we did not receive the email" is in fact a valid
excuse if the email in question quotes spam. I didn't have alternate
contacts listed, but I could have listed a dozen and they would most
likely all have classified Andy's email (10% "what is this" and 90%
quoted spam) as spam.
There's always pastebin...
It has an API so sending these spam reports can be automated.
--GM
27 Dec
27 Dec
11:53 p.m.
b) After at least 21 days of sending email alerts to
the main
contact and the emergency contact and receiving no response, the
VPS's networking will be suspended. Networking will be re-enabled
when contact is re-established and a plan for securing the
problem service is agreed by both BitFolk and the customer.
I have already broached this question on IRC and basically no one
was in favour of (a) because they did not feel that surprise
invoices would go down well.
I actually got a problem with b).
My home (personal) contact stuff goes almost exclusively via bitfolk.
That includes email/sms/jabber/phone[voip], so if my VPS gets shut down
it's difficult to contact me. (There is some redundancy but not in the
frontend really and I can't be bothered to create automatic fail-overs
etc..)
When there are problems and I don't respond, it's usually because I'm
not receiving the messages and are thus not in a position to do anything
about it.
Cutting off the server makes the matter worse for me.
For me, keeping the service going is top-most priority.
Could I "elect" to to be invoiced rather than shutdown? For example, as
an option on the Panel?
4 Jan
4 Jan
9:19 p.m.
Hi Conrad,
Rearranging your email a bit…
On Wed, Dec 28, 2016 at 12:53:36AM +0100, Conrad Wood wrote:
Could I "elect" to to be invoiced rather
than shutdown? For example, as
an option on the Panel?
I've been thinking about it for the last week and I'm going to have
to say, "yes YOU can, but in general the answer's going to have to
be no."
See the thing is, I think it's very likely that if I was to write up
this alternate option in the terms, and add the necessary check box
etc to the Panel then the only person who'd use it is you. So it
would represent a lot of confusion and complexity for very limited
value.
However, I am happy to make a note in our customer database that
should there be some problem with your VPS that we need you to fix
and we are unable to contact you for at least 21 days, then instead
of turning off your VPS's network we'll add a firewall rule and
invoice you for it on a monthly basis. Assuming that there *is* a
firewall rule that we can add that fixes the problem.
When there are problems and I don't respond,
it's usually because I'm
not receiving the messages and are thus not in a position to do anything
about it.
I suppose I'm a little surprised that we could email you daily for
weeks and then send you text messages and you still would have no
ability to read them and rectify whatever the problem was. All the
things we've discussed in this thread have been things you can solve
with a firewall rule and/or a simple application reconfiguration.
I'd also like to think that we only go to these lengths about the
small number of things that are fairly serious risks. Things you'd
really want to fix quite quickly if you knew about them.
If the problem would be that you'd not be receiving the messages (as
opposed to receiving them but not having opportunity to act on them)
is there some other communications method that would be more likely
to get through than email and text message?
If so then I could certainly look at adding that to the address book.
For me, keeping the service going is top-most
priority.
I really don't want to be turning anyone's service off either, but
sometimes we do need to get people to do things. A year ago we only
had one email address for each customer so when daily emails to that
address went unanswered for months at a time we were a bit stuck as
to what to do.
Since then, the ability to provide multiple email addresses and also
phone numbers was added, so now we have more options, and before
turning anyone's service off I'd very likely exhaust all available
contact methods¹. Now that can happen, I feel it's a reasonable time
to enforce some limits on how long we'll wait.
If there's other contact methods that would work better for some
people then I am keen to hear about them, to see if we could make
use of them.
Cheers,
Andy
¹ Except in emergencies, obviously. If there's an abuse report, and
we look at traffic and see abuse actually happening then it just
has to be turned off without time to send email or whatever and
wait for response.
--
https://bitfolk.com/ -- No-nonsense VPS hosting
2908
days inactive
2928
days old
28 comments
14 participants
participants (14)
-
Adam Spiers -
Alun -
Andy Smith -
C. Scott Ananian -
Conrad Wood -
G. Miliotis -
Matt Molyneaux -
Michael Stevens -
Ole-Morten Duesund -
Paul Lewis -
Richard Dignall -
Robert Gauld -
Robert McWilliam -
Roger Light