Fastmail & POBox - Elegant redundancy - NO MORE
 

I have been a user of both Fastmail & POBox for very many years.

I like FM as a service provider & POBox historically provided elegant redundancy.

However today's outage:-
http://www.emaildiscussions.com/showthread.php?t=73859
proves that this is no longer the case, and that there is one, or I fear more, single points of failure.

May I suggest to FM management that the two services should be run from different data centres, in different continents, as they were historically, so as to reinstate the elegant redundancy we enjoyed in the past.

NJSS

Yes they use to have a couple of back up servers in different locations.

maybe it was before ^^ and they $topp€d du€ to................................................................

vivil

If I recall correctly this stopped because of the merger of FM & POBox.

NJSS

whatever the reason, it is abnormal ^^ DC hardware have redundant connections, energy supply

The big concern I have is that there is only a vague description of the problem. There is no way of judging how likely the issue is to recur. We are not even sure the merger of FastMail and POBox was the reason both went offline, though this seems likely.

Even those of us that like FastMail, if mail outages of more than a few minutes cannot be tolerated, need to think about a backup mail with one of the major providers. Fortunately, FastMail still has capabilities that make this easy to set up, assuming not all MX servers are affected by outages.

For my own domains I have the MX records pointing as follows:

I was confident that Fastmail would get their servers back up but of course the timings were not known at the time.

Can anyone suggest an alternative provider that I can have as a backup whilst Fastmail MX servers are down such that I can still receive and respond to email. For example will googlemail allow me to use them in this way?

But I guess, taking into consideration my SPF and DMARC records, a third party MX server will make things complicated.

I looked into mixing MX servers from different providers at one point. In theory, it seems like a good idea. Unfortunately, the priority settings do not seem to guarantee that he servers will be polled in that order. It might work if you carefully ensure there is two-way syncing between your two mail providers, but it is non trivial to get everything right. FastMail supports immediate forwarding of a copy to your backup mail provider at MX time, but most other providers do it less cleanly. Syncing outgoing mail is sometimes also an issue, assuming auto bcc is not supported.

Thanks BritTom

Yes it does seem quite complicated and not sure an external tertiary MX server will work anyway. As I understand things the secondary MX server Fastmail uses simply will take incoming mail and cache it until such time as the primary MX server gets back online. But not clear if the secondary MX server was down in today's outage. Hopefully Fastmail's secondary server is in a different location and uses different providers? I guess it must be as otherwise Fastmail can't guarantee no lost mail.

Perhaps one way is to set one's MX records such that the primary points to in1 at fastmail, the secondary to a 3rd party MX server and a tertiary to in2 at Fastmail? Would that work?

In your research did you come up with a provider you could use?

FastMail naps amid network blunder
 

Unfortunately, according to Pingdom, the secondary MX server (in Seattle) was down for about an hour. This was disturbing, because the backup email solution I use for a couple of customers relies on the fact that, while incoming mail is cached at the MX server for later delivery if FastMail main servers are unavailable, the MX server arranges for immediate delivery to additional locations. That is, an alias of xxxxx@f-m.fm which specifies delivery to xxxxx@fastmail.com and xxxxx@gmail.com will queue the messages while FastMail is down but deliver immediately to Gmail. When all MX servers are down, that does not work.

SMTP concept (or well-known reference sendmail software concept) is to use the lower (or higher i forget) number server before the DNS MX request and using for the second try another server and so on...

Of course, the individual CPU/memory/network of every of these MX ***must*** be twice
the maximum capability daily e-mail volume of these MX to pick-up collapsed traffic from down MX servers..

RUNBOX use only one MX (ok its TTL is 5 minutes but...). I quit them for this reason; i did not know if they has a secondary MX backup...

Hmmmm...........

https://www.google.com/search?=fastmail.com+dnsinspect

give us the january 2018 cached page which contains the same result as now:

Reverse Entries for MX records
OK. All mail servers have reverse DNS entries configured correctly.
Server IP PTR (Reverse) IPs
in1-smtp.messagingengine.com. 66.111.4.72 mx3.messagingengine.com. 66.111.4.72
in1-smtp.messagingengine.com. 66.111.4.75 mx6.messagingengine.com. 66.111.4.75
in1-smtp.messagingengine.com. 66.111.4.73 mx4.messagingengine.com. 66.111.4.73
in1-smtp.messagingengine.com. 66.111.4.74 mx5.messagingengine.com. 66.111.4.74
in1-smtp.messagingengine.com. 66.111.4.71 mx2.messagingengine.com. 66.111.4.71
in1-smtp.messagingengine.com. 66.111.4.70 mx1.messagingengine.com. 66.111.4.70
in2-smtp.messagingengine.com. 185.68.180.24 smx2.messagingengine.com. 185.68.180.24
in2-smtp.messagingengine.com. 185.68.180.20 smx1.messagingengine.com. 185.68.180.20

https://www.maxmind.com/fr/geoip-demo gives US an NL

so... 6 internal MX servers is on a network 1 and the last 2 are on network 2....

one) the 2 networks links collapsed in same time during about 1 hour.... bad luck !

two) if these 8 MX have the same CPU/memory/network links with a weak "margins": if network 1 collapsed more than 18h (6/8 of 24h) with 0 troubles in network 2... lost mail can occurs.

still a big silence from FM staff about the small shutdown of the 2 main MX servers during this outage ^^

Can not make out of your post, if you’d asked them for an explanation (FM is not coming here often anymore).