One of the associations I'm in (besides LUV) have been experiencing
difficulties sending bulk emails to members on a monthly basis (for
announcements, newsletters purposes). This is because when you are trying
to send bulk emails via your free web mail account (in this case, GMail),
there is a limit imposed by Google (or any other webmail service providers)
on the number of email addresses we can mail out to, within a period of
time - and this limit can get ever stricter as time goes by. So the whole
mailout process becomes unmanageable after a while.
While the Ctte is looking into various alternatives, including several of
the 3rd party mail services / transactional mail services in the market, I
am offering them another option: Setup a SMTP mail server (e.g. Postfix)
on a VPS hosted by a web hosting company, and handle the whole bulk mailout
ourselves. I am willing to help them setting up a VPS, hardening the
Linux, installing & configuring all the mail software, etc, and then take
care of the regular maintenance.
For VPS hosting service - I had looked at one hostwinds.com (Editor's
Choice of au.pcmag.com 2017) (~ US$ 14 / mth). Will check out others to
compare - basically I'm after a reasonably priced and reputable VPS web
hosting company. I would like to hear any recommendation, or any
first-hand experience dealing with any particular web hosting company -
which I'm sure many of LUV members certainly have.
For domain name (.org.au) - About $48 for 2 years. Any reputable Domain
Registrar that you would recommend?
For SSL Certificate - I can install one from Let's Encrypt, which will be
free.
Any major cost/charges that I have missed?
I also intend to follow all the bulk mail best practices, SPF/DKIM
settings, MX records, security measures, email etiquette, etc, to make sure
that our organisation normal notification/newsletter to our members, would
not be treated as SPAM.
The 3rd party mail service providers that I had found so far: MailChimp,
MailJet, Amazon-SES, SendGrid, MailGun (and many others).
Did lots of google searches, trying to compare (Advantages & Disadvantages)
between managing own SMTP mail server and using 3rd party mail services.
The results were mix and varied.
I myself certainly have much more faith in opting for a self-managed mail
server solution over the commercial mail service ones.
Some of the advantages:
- Have full control over most aspect of the mailing process. More
flexible. Not constraint by the ever-changing rules imposed by 3rd party
mail services.
- Using all free & open source software - have all the freedom, and can
tap into the large FOSS Community out there for various kinds of help and
support.
- Cost-wise, the mail server option is generally on the lower end of the
price spectrum, as compared to many of the 3rd party mail services in the
market.
- No doubt the own mail server option would involve more work, more time
to manage the whole thing - but I'm quite happy to take up this challenge,
and to learn a lot from this hands-on experience.
I would be delighted if any of you who are experienced in managing a
linux-based mail server can kindly share with us some more real-life
examples and hands-on experiences - so that I can gather more solid
evidence & cases, hopefully to strengthen my case for a self-managed mail
server solution.
Some points that I had read somewhere that I would like to quote below to
seek your comment:
"*What a commercial mailing list operator does, and which you cannot easily
replicate, is build and maintain a reputation as a responsible and
reputable source of bulk email. If you are not going to be getting into
that, perhaps you could benefit from getting help from a commercial mailer.
... *
*Abuse handling, bounce handling etc are important to get right, but until
you have significant experience, your reputation is going to remain zero,
if not negative (which is a reasonable starting point for unknown domains
in this day and age). A lot of the reputation-based stuff like SPF, DKIM
etc will help only if you have a reputation to defend. On the other hand,
it does send the right signals to somebody who is deciding on whether or
not a sender is to be treated as reputable (or rather, their absence is not
a good sign; neither, in my book, is anything to suggest you are using
homegrown and/or prerelease software to send email). *..."
I also read somewhere which seemed to point to a 3rd option: Still use own
mail server, but direct the outbound mails to an external Mail Relay
service. What do you think?
When this whole project (if accepted by my Ctte) is done and implemented
successfully, I intend to share my experience in this whole process, in one
of LUV talks.
In the mean time, I have a lot of work to do, and lots to learn ...
Cheers,
Wen
Recieved-SPF:
permerror (darrenwurf.com: Unknown mechanism type 'aaaa' in 'v=spf1'
record)
receiver=luv.asn.au; identity=mailfrom;
envelope-from="darren(a)darrenwurf.com";
helo=mail.darrenwurf.com; client-ip=43.229.63.47
https://vamsoft.com/support/tools/spf-policy-tester
46.4.124.163
darren(a)darrenwurf.com
PARAMETERS
DNS server: 8.8.8.8 (Google Public DNS)
Evaluation time limit: 20 seconds (default, see RFC7208 Section 4.6.4)
Maximum number of void DNS lookups: 2 (default, see RFC7208 Section
4.6.4)
Standards compliance: RFC7208 (April 2014)
Test is queued, please wait...
+0 msSPF check starting.
IP: 46.4.124.163
Sender: darren(a)darrenwurf.com
Domain: darrenwurf.com
EHLO/HELO domain:
+0 msRetrieving DNS TXT record for "darrenwurf.com".
+37 msTXT record found.
Line #1: "v=spf1 mx mx:darrenwurf.commx:secure.darrenwurf.commx:smtp.darrenwurf.coma:darrenwurf.coma:docker4.darrenwurf.comaaaa:darrenwurf.com -all"
+0 msStarting SPF policy evaluation.
Policy: "v=spf1 mx mx:darrenwurf.commx:secure.darrenwurf.commx:smtp.darrenwurf.coma:darrenwurf.coma:docker4.darrenwurf.comaaaa:darrenwurf.com -all"
+1 msSyntax error in the SPF policy. Returning SPF "permerror".
Domain: darrenwurf.com
At character: 119
Length: 19
Token: "aaaa:darrenwurf.com"
Error: Unknown term "aaaa".
+0 msFinished SPF policy evaluation.
DNS limits status: DNS terms 0 of 10 allowed. Void lookups 0 of 2
allowed. See RFC7208 Section 4.6.4.
+0 msThere was an error during the policy evaluation.
Error message: "Syntax error in SPF policy for "darrenwurf.com" at
character 119: Unknown term "aaaa"."
+0 msPolicy evaluation finished with SPF "permerror".
TEST SUMMARY
The evaluation completed in 38 ms, with 2 errors and 0 warning.
Result: SPF permerror
Permanent error. Fixing the problem requires intervention of the
policy publisher.
Error message: "Syntax error in SPF policy for "darrenwurf.com" at
character 119: Unknown term "aaaa"."
Hi Arjen,
Thanks for your info.
I will check out this Binary Lane. BTW, I noticed a few of our LUV members
have VPS hosted by Digital Ocean. They seem quite active in hosting various
Linux related forums and tutorials. However, Digital Ocean is also
American, but they do have DCs outside of US, e.g. Canada.
Cheers,
Wen
On 23 Sep. 2017 5:00 pm, "Arjen Lentz" <arjen(a)lentz.com.au> wrote:
Hi
For our VPS hosting needs in Australia we use Binary Lane. They're
Australian owned as well, which I regard as an advantage (also in legal
terms).
Stuff just works and their support has been very prompt and technically
competent.
Linode is very good and we've used them for years, but they are American
and thus directly subject to US legal leverage. Even if you used a Linode
DC in Japan or Singapore, ultimately it would be open to US legal
frivolities.
This may not be a consideration for some needs, but I'm just putting it out
there.
Data sovereignty is a thing.
Regards,
Arjen.
On 08.09.17 22:22, Craig Sanders via luv-main wrote:
> except that udev will **always* run before any NIC is up, while ifrename
> may not - and will bail if the NIC is in use. which, IIRC, provided the
> motivation for me to finally switch from ifrename to udev years after
> udev had the capability.
Whatever works best in the use case, of course, but from "man ifrename":
"Ifrename must be run before interfaces are brought up, which is why
it's mostly useful in various scripts (init, hotplug) but is seldom used
directly by the user."
Sounds like mdev will be worth a look once it matures a bit. (Will have
to try to remember.)
Erik
On Thursday, 7 September 2017 11:02:52 PM AEST Rick Moen via luv-main wrote:
> I would never recommend for a business as file server with simultaneous
> use of motherboard SATA ports, a PCI-E SAS card, and USB things on an
> ongoing basis. That seems like poor component selection, IMVAO. [0]
Sometimes you just have to do such things.
I run a server which has as it's main purpose preparing SD card images for
embedded PCs. The images are made by the people who install the embedded PCs
on-site (who wear high-visibility clothing).
Getting them to unplug USB devices during system boot isn't a viable option.
I just have to make the system work with a SD card being sda and hard drives
being sde and sdf if it boots with the USB device connected.
As for Ethernet device names, here are the device names I chose for one of my
systems:
ethbl
ethbr
gethm
getht
lo
mb0
The sockets on PCI boards are bottom-left, bottom-right, middle (gig-e), and
top (gig-e). The motherboard has a single socket. Udev naming worked for me
and didn't need any special configuration. Now I use files in /etc/systemd/
network and again it works without any special configuration.
--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/
I just remembered:
> I vaguely recall that a system swapped eth0 and eth1 when replacing a
> 2.0.x kernel with a 2.2.x kernel (or 2.4 to 2.6, or something like
> that). Which didn't surprise me much, and is why God made rc files
> editable.
>
> And ifrename is cool.
I've encountered _zero_ instances of network interfaces changing their
device nodes on RHEL/CentOS, under any circumstances, for the simple
reason of DEVICE and HWADDR directives being used in the default
/etc/sysconfig/networking/devices/ifcfg-* files.
DEVICE=eth0
HWADDR=11:22:33:44:55:66
Which can optionally be used to assign devices names of your choosing
such as 'lan' and 'wan'.
I guess the 'network interfaces have come up in a different order' scenario
seems to be primarily a Debian/*buntu, etc. one (that has never arisen
in my use-cases), and mostly involves USB (and other equally flaky
hotplug hardware schemes).
Did I mention that ifrename is cool? ;->
Success.... At last!
I was getting no where on Debian 8, glancing at Debian 9 spec, it said
it had support for G4 modems, so I install debian 9, with network
mangager..............still no joy .........ssssssiiiiggghhhhhh.
Anyway this morning I was ready for a challaenge so I decided to have
another go.
I ran the command "nmcli connection show" to see what was happening and
was stunned to find the following report...........
----------------------------------------------------------------
enx0c5b8f279a64: connected to Wired connection 3
"HUAWEI MOBILE"
ethernet (rndis_host), 0C:5B:8F:27:9A:64, hw, mtu 1500
ip4 default
inet4 192.168.8.100/24
inet6 fe80::4636:a274:6574:96fa/64
enp5s0: unavailable
"Marvell 88E8056 PCI-E Gigabit Ethernet Controller
(Motherboard)"
ethernet (sky2), 20:CF:30:7E:C3:09, hw, mtu 1500
lo: unmanaged
loopback (unknown), 00:00:00:00:00:00, sw, mtu 65536
DNS configuration:
servers: 192.168.8.1
interface: enx0c5b8f279a64
servers: 192.168.254.1
domains: mydomain.example
interface: enp6s0
---------------------------------------------------------
Its enx0c5b8f279a64 that I am looking for, it now works OK,
interestingly connection time is slower than with windows but transfer
rates appear to be good.
Why I did not do this before, god knows............
God its great to getaway from bl...y awfull windows.................
Now all (ALL!!!) I have to do is to reconfig my fire wall on to this box
and my small network is back on the air.
Lindsay
The luv server was down this morning because of a KVM error. Also another KVM
VM on the same system crashed. Sorry for sleeping in.
--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/
There are meetups for hacking open-data type projects held in the CBD. They
are quite ok and hacking any open software is acceptable (some people edit
Wikipedia and I have done some Debian development there). But they aren't
specific to Linux and similar development.
I have access to a very comfortable office in the CBD with free tea, coffee,
hot chocolate, and wifi and I can invite friends for meetings. I can't issue
an open invitation, but probably everyone who would be interested in attending
such meetings is a friend anyway. ;)
Who would be interested in attending such a meeting to work on Debian and
similar projects? People who want to test software, file bug reports, work on
documentation, etc are welcome.
--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/