Are you bloody joking ?

[df55555]

The account you are trying to login to is on a server which requires emergency maintenance. The maintenance will take around 2 days - it should be complete by the end of the weekend. Update will be posted to FastMail.FM's status page

[injinuity]

2 days ..ouch.. but I take solace in the fact that this is the first time something of this magnitude has happened in around 4 years... and I have a feeling it might not again.. so lets just keep our fingers crossed and hope for the best.

Jinu Johnson

[TedCavitt]

but I take solace in the fact that this is the first time something of this magnitude has happened in around 4 years

WRONG WRONG WRONG. I distinctly remember several week-long outages over the past three years, including the database corruption and the power outage. Total joke.

[injinuity]

errr *of this magnitude* as in over 10 hours.. as in over a day.. this is the first time it is actually expected to span over a day...

[df55555]

I just had to mail a whole boatload of contacts - saying use this address today please. Then reply to the majority of them to explain why - then reply again explaining why I recomended they use fastmail for email.

I have only renewed this account annually as I managed to get my first_name@ - ffs spam/sieve et all is all good - but reliability trumps all those hands down.

In additiion to that Thunderbird keeps trying to log into fastmail despite being told explicitly not too.

I've ranted enough for now, I'm too depressed to read the larger thread on this issue. I will most deffinately be reconsidering my account here and the one I bloody created two days ago for a friend(salt into the wound that)

I might just go out and get a large fried breakfast from my local cafe.

[FastGrimm]

Emergency maintenance can happen. But up to 2 days to restore maildatabases of paying customers is just too long.
If the problem was with a free emailservice such as hotmail or gmail I could understand, because since those services are free they might have a lower customer service level. Not sure though, but at least I wouldn't depend on them so solely as I do on Fastmail.

I see a serious single point of failure here, and that's the fact that a user's mailbox is stored on a unique server. In the case of anything going wrong with that particular server all mailboxes thereon are not available. The result in this case:
Only the paying customers are concerned. The free accounts are all up and running.
Notice how weird that sounds to the untrained ear?

Why isn't something like a shared filesystem on clustered fileservers an option? Why did you choose a setup that sounds so amateuristic to many people? How did the filesystem got corrupted for example? Why does the restore process take so long? How do moguls like Hotmail, Yahoo and GMail have this handled? In my opinion your recoveryplan in case of a disaster (I think we can call it that by now) of this scale is simply not sufficient enough.

These kind of events simply require me to rearrange the way my mail is delivered and keep a local backup of my mail at Fastmail. (I looked around and think ImapSize (free) might be a good one for that)

To anyone who has something to say in the operational management of Fastmail:
An outage of up to 2 days is not an option.

If it is for you, be fair enough to let us know.

[InResponse]

Originally posted by FastGrimm

These kind of events simply require me to rearrange the way my mail is delivered ...

Interested, obviously - what do you have in mind?

[EdGillett]

FastMail boys (that means you Jeremy, Rob, Bron) - This is bad.

Incidentally since the last outage you had in July (not half as bad, but still ...) it's been fine for me.

What concerns me is at the time, we were told in this post by Bron (http://www.emaildiscussions.com/...764#post327764)

Some key qutoes from Bron in that post:

"To improve stability and performance: Two new 4TB arrays in the past month. We're working towards installing a much more efficient POP and IMAP proxy, alleviating the load on our frontends.

At the moment, 50% of our servers are critical - meaning that we can't shut any of them down without users noticing a degredation of service. We're working with our hosting provider to bring that number down, and working with the Cyrus development team to migrate to a mirrored arrangement on a mail storage servers so we can restart one of them without mail downtime for the users hosted there"

OK - so - what happened to the redundancy plan?

It's evident that your infrastructure is still not able to cope with the loss of a single server.

The fact that we are looking at such a long period of downtime is going to be your undoing on this one.

I understand that issues happen - it's the nature of server s and computers generally to start complaining for no apparent reason.

However, If I turned round to one of my companies clients and told them their system would be unavailable for 48 hours, we would be fired on the spot. It's simply not acceptable.

Fine, fine - I know there's a lot of data to restore off backup tapes and I know how fast these things can physically run and probably in this situation there is nothing more that you can do to get service restored.

But wasn't it situations like this which should have been taken care of back in July when you told us you were working to improve redundancy?

All the features in the world, all the community of users you have built up around you which remain one of your strongest aspects, mean nothing when you're blown off the face of the web for DAYS.

Your servers should be mirrored. Our data should be stored on redundant storage arrays. Downtime should be minimal.

Corrupt file system? What at the OS level?

Your users need to know what you are going to do to resolve this failure in your infrastructure and when it's going to happen.

I too am now going to have to speak to all the people I recommended FastMail to who are wondering when their mail will come back.

One of which is a business customer, who perhaps, in hindsight should have been hosted with an email host with an SLA guaranteeing uptime.

I have to say that at this stage I will not be able to recommend FastMail for business use until you've corrected deficiencies in your infrastructure.

I know you're doing what you can to resolve the current situation swiftly, and your efforts are appreciated, but I fear this time an outage of this magnitude will have shaken the confidences of even your most long standing clients - not because you went down, but you went down AGAIN and the work we were promised to improve redundancy simply hasn't ocurred.

Does FastMail have a formal SLA in place for paid members?

Thankfully I have my own domain. This means that I can move where anything@mydomain.com gets shifted to reasonably quickly. Unfortunately at the minute, I have my MX records pointed at FastMail, and this is going to change.

I'm going to need to move my MX records away from FastMail and instead set up forwards from my web host to FastMail (My webhost is 1&1, who I've been using as a web host for my own use for about 6 months now. And they're bulletproof because they're 1&1 - their data centres are the most redundantly configured anywhere in Europe.) And you're damn sure that if there was a really serious outage which somehow managed to take them offline, it wouldn't be anywhere near 48 hours.

I just can't rely on FastMail's reliability, and in order to claw back some control, I'm going to need to externalise my domain and rely on forwards from now on.

[self]

Even if it would seem that "messagingengine.com" is still operational and can forward mail to backup destinations/handlers as previously configured (with a delay though), this is not at all acceptable.

[EdGillett]

I have a paid Enhanced account, and have been a happy customer over the years, but recently, FastMail is not cutting it on the reliability front. This recent incident has meant I've re-evaluated my email needs and moved accounts.

1&1 offer 1GB IMAP mail accounts, with full domain management. From 69p a month for a 1GB IMAP Account! And that's hosted on 1&1's infrastructure.

It's not fair to compare the resources which 1&1 command to FastMail. FastMail have been doing a sterling job until recently, but looks like they don't have the resources to provide the uptime required for BUSINESS CRITICAL email.

For anyone seriously concerned with uptime on ANY email provider you should setup your email to be as redundant as possible.

Firstly - control your own email address - get your own domain. You can do this very cheaply (£1.99 for .co.uk domains here in the UK through 1&1)

Once you've got your own domain (me@myname.com) you can move your email account to whoever you desire, but retain the same email address wherever you go.

Now - you can forward this me@myname.com to multiple places as well. Rather than going to one place, I now have my primary email sent to My FastMail account, a Gmail account, and also one of the 1GB IMAP Accounts I have with 1&1.

Concerned with server uptime - have a look and see why 1&1 deserve their reputation - look at what they're running: http://order.1and1.co.uk/xml/order/AboutUs or http://www.1and1.co.uk and browse to About us.

FastMail is great, but it's proven itself to be unreliable of late, and sad though I am to say it, it's no longer my primary mail account, and I will be migrating my data across to 1&1 over the next month (As soon as I can get access to it again.)

A sad day indeed ...

I'll keep up to date with the forums and the status blog though.

FastMail guys - I encourage you to post a statement on the status blog explaining what you intend to do about your reliability and establishing server infrastructure redundancy.

[FastGrimm]

Originally posted by InResponse
Interested, obviously - what do you have in mind?

Well, all mail I get is sent to a couple of domains I have, not directly to my Fastmail address. These domains currently all have Fastmail's dns-records. I can change them to default records and for example set up a forwarder that forwards all mail to two mailproviders instead of one, so every mail gets delivered in my inbox at Fastmail and my inbox at Gmail.
Downside is that I'll have twice the amount of mail to sort out

[EdGillett]

My primary account is sorted and I maintain a decent folder structure there.

The others are there to collect email for use in emergencies - Gmail is useful since it's free and offers buckets of storage and the user interface is designed for quick searching. I never use it until I need to access mail in an emergency.

Anywho - hopefully my new primary email will be more reliable.

I'm disappointed with the fact that the 1&1 system doesn't allow you to setup any sorting rules server side - this was a major bonus of FastMail.

I may still end up using FastMail as an email store, but I've moved MX records to my web host and forwarded there now.

I'm arranging for business users I've recommended to use FastMail are moved onto a provider with an SLA in place (such as 1&1's 99.9%).

[rstmaniax]

Currently in the process of moving the 3 companies i recomended to fastmail somewhere else aswell as my own.

Also a dozen or so personal accounts of friends/family will also be moving.

This is beyond a joke.

[allpar]

Hate to say it, but I agree. I have my own server and may just start looking for seive replacements.

[fahrner]

Originally posted by EdGillett
Gmail is useful since it's free and offers buckets of storage and the user interface is designed for quick searching. I never use it until I need to access mail in an emergency.

I have often thought about such a backup account.
How do you mange this account? When you send copies of each mail to your gmail account, how do you cleanup your inbox at gmail? When you need it some time, you will find hundreds of mails in your inbox, if you do not clean it regularly.

Jochen