?
Or just buy a Synology Disk station.
A project that aims to turn the Raspberry Pi into a personal web and mail server has stuck out its hand for crowdsourced cash so it can finish off its platform. ArkOS, for that is the service in question, promises a cut of Arch Linux customised so that if offers an easy-to-use web server and mail server capable of running on a …
I wouldn't trust my email to an SD card, since playing with Sheevaplugs and Pis, I've had several cases of corruption on SD cards, including one in the past week. However, I run my mail server on a Sheevaplug with an external hard disk for more reliable storage. That's the one where the SD card went dodgy, but a quick swap out for a new one and it's all up and running again. I need to look at the configuration a bit more closely to eliminate as many writes to the SD card as possible.
As with anything, provided you've got backups of important data, it's recoverable. The only reason cloud providers are safe (and then not always) is because they do have some redundancy in their system. at least with your own personal IMAP server, you've got full control over your own email, and if it's a small, low-powered device, it can be left on and configured so you can access it remotely from your smartphone.
> I need to look at the configuration a bit more closely to eliminate as many writes to the SD card as possible.
Yes, I've discovered the same problem. I have a Pi in a remote location (my Mum's house) that' is running 24*7. It used to burn through SD cards in a couple of months in normal operation. I diddled around a bit and put /var/log and /tmp on tmpfs . So far this card has been running for 6 weeks and no obvious corruptions yet. Fingers crossed.
But I'd never trust the Pi as the sole storage device for any valuable data. I don't even trust that it will run for months or years unattended.
Errrm guys, just put your O/S partition on a USB drive, I had the same issue with the Pi and SD cards, moved it over to an external USB hard drive and away you go (preferably choose a powered USB drive, otherwise make sure the Pi is getting power direct from a good power source, and run the HDD from a powered hub otherwise you'll get phantom bad sectors (which don't actually exist) due to low power.
Plenty of articles on the web on how to do this, all you have to do is amend one entry in a config file on the FAT partition that starts the PI up, after doing this the SD card is only read at startup, and is easily replaced if it goes bad.
> You need an SD card made with SLC Flash
You could well be right. But doesn't that just delay the inevitable failures?
I appreciate that the Pi was never designed, nor meant, to be used in environments where reliability was important, but there doesn't seem to have been much work done with the Linux distros to mitigate what must be a very common failure mode.
I need to look at the configuration a bit more closely to eliminate as many writes to the SD card as possible.
SD cards should not "burn through" in a couple of months. The one in my Sheevaplug has been fairly hammered 24 x 7 for a couple of years. I have a Pi whose SD card gets file system corruption occasionally, but that doesn't necessarily mean the card is at fault. I pop it into a Linux laptop, fsck -a and put it back. No problem. Never had an SD card that actually died.
Linux check 3.6.11+ #371 PREEMPT Thu Feb 7 16:31:35 GMT 2013 armv6l
Welcome to check Unix
Last login: Mon Nov 4 16:01:55 2013 from 192.231.21.245
nick@check ~ $ uptime
15:44:00 up 212 days, 20:58, 1 user, load average: 0.12, 0.38, 0.29
nick@check ~ $ top
top - 15:44:08 up 212 days, 20:58, 1 user, load average: 0.10, 0.37, 0.29
Tasks: 86 total, 1 running, 85 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.7 us, 1.3 sy, 0.0 ni, 98.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 481288 total, 449468 used, 31820 free, 72548 buffers
KiB Swap: 102396 total, 0 used, 102396 free, 282804 cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
16825 nick 20 0 4664 1364 1024 R 1.0 0.3 0:00.10 top
37 root 20 0 0 0 0 S 0.3 0.0 1096:12 mmcqd/0
1 root 20 0 2140 608 516 S 0.0 0.1 10:27.76 init
2 root 20 0 0 0 0 S 0.0 0.0 0:06.17 kthreadd
3 root 20 0 0 0 0 S 0.0 0.0 10:07.40 ksoftirqd/0
5 root 0 -20 0 0 0 S 0.0 0.0 0:00.00 kworker/0:0H
7 root 0 -20 0 0 0 S 0.0 0.0 0:00.00 kworker/u:0H
8 root 0 -20 0 0 0 S 0.0 0.0 0:00.01 khelper
9 root 20 0 0 0 0 S 0.0 0.0 0:00.00 kdevtmpfs
10 root 0 -20 0 0 0 S 0.0 0.0 0:00.00 netns
12 root 20 0 0 0 0 S 0.0 0.0 0:03.48 bdi-default
13 root 0 -20 0 0 0 S 0.0 0.0 0:00.00 kblockd
14 root 20 0 0 0 0 S 0.0 0.0 0:00.22 khubd
15 root 0 -20 0 0 0 S 0.0 0.0 0:00.00 rpciod
16 root 20 0 0 0 0 S 0.0 0.0 0:19.07 khungtaskd
17 root 20 0 0 0 0 S 0.0 0.0 0:11.12 kswapd0
18 root 20 0 0 0 0 S 0.0 0.0 0:00.01 fsnotify_mark
arkOS has a tool to buffer log writes in memory that eases SD card wear considerably. Also I am planning on making it easier to store data on connected USB devices, because SD cards are really not great for stable long-term storage.
Problems like this are well within the scope of the project and definitely things I'm thinking about. :)
Thanks!
This isn't what I would call crowdfunding it looks suspiciously like soliciting donations for the personal benefit of one developer for something that is supposedly developed by the community. There are no details on what the funds are required for or would be spent on and there are no goals or timelines. Although there is a target the way it is setup means the money is taken even if the target isn't reached.
That was my thoughts exactly. This is basically asking for twelve month's salary for a bit of meta-packaging and a certain amount of PHP/Javascript coding. There's no way to know precisely how much work is anticipated there but you could do a hell of a lot in a couple of weeks, certainly enough to cover the common set of tweakables.
This seems to me to be purely an exercise in personal enrichment. There's nothing wrong with that per se, but it's not really a "community" project if that is the case.
"This is basically asking for twelve month's salary for a bit of meta-packaging and a certain amount of PHP/Javascript coding. There's no way to know precisely how much work is anticipated there but you could do a hell of a lot in a couple of weeks, certainly enough to cover the common set of tweakables."
It's actually written in Python but I suppose the real point still stands - namely it doesn't need a lot of work to get something reasonably feature-complete. This has been around for a while and purely out of interest I got a copy to have a look at and I have to say I'm underwhelmed. The very first file I looked at was the Genesis Makefile to see how everything fits together: it's full of those lone-developer assumptions that should have been bounced out long ago for a project that has got to the point of needing public support. For example it assumes it will be built directly in the user's home directory: has ANYONE else tried building this code? If so why is that still there?
I then looked through the actual source code. I didn't go through it in fine detail but it simply doesn't LOOK like good Python. Dynamically typed languages in particular need a very defensive mindset if non-trivial apps are to be sufficiently robust, but you don't see anything of that kind in the code. You might be able to get it working but it isn't even the beginnings of professional grade code.
Yes, this smells bad and I can't help but feel people are going to go away disappointed.
"'pacman -S postfix dovecot roundcubemail'..."
Quite - I was htinking of Citadel, which would do it all. Inevitable that there'd be piranhas snapping around the Pi's appeal, though. Hopefully the fact that the Pi encourages people to learn for themselves will lead Pi owners to finding out that it's all there for them already.
The funds being raised are so that I can work on the project full-time, thus drive its production forward. I never make the claim that the project is entirely developed by a community, rather that I am the full-time developer and others are contributing some things to the project when able. The goal is to have a stable version of the software for general use by March of next year, then over the course of the year to build in the features that are outlined on the page, as it says. The setup is analogous to a "Flexible Funding" campaign, the likes of which you would see on IndieGogo. Nothing new here.
I think if someone's technical enough to want to run their own web/mail servers, I'm sure they are capable of setting up a port forward in their router. There's also that dirty nasty security hole that is UPnP which I imagine most non-technical people have left enabled.
"The days where ISPs hand out a _full_ IPv4 address to every customer are gone."
Really? I seem to have a _full_ IPv4 address (I presume by full you mean non-NAT'd at the ISP). To my knowledge, the only ISPs doing that are mobile carriers, which only an insane person would host a server over.
This seems really fishy. And pointless. Anyone interested in this sort of thing could easily set up their own server on a pi following a few tutorials off t'web.
I especially don't like how they've made their funding area look like it's a kickstarter campaign when it isn't. Seems there's nothing to stop them taking funds even if it doesn't meet its whopping 45K target.
I say 'they' - it appears to just be some 23 year old. Suuuuuuure he can be trusted to properly manage a 45k budget project.
In the words of Dragon's Den - "thanks, but I'm out".
There are enough very simple distros out there to do this.
This is really a setup task rather than a product.
Just download any linux distro, follow the setup for its apache component and download iredmail, a mix of other open sources apps that installs a functioning mail server in 5 minutes. Its not the best mail server out there and is really just a marketing technique for their other products but it works for small scale mail server.
To be honest, you really should install this stuff manually so that you know how to handle linux and the other software. Otherwise you can end up with massive security holes or major failures.
As for IPv4, its actually pretty simple. You just use dynddns, most ISPs router firmware supports dyn's login making a dynamic IP no problem. You just have to use your ISP to send mail as most mail servers treat mail from dynamic IPs as spam.
"Any Raspi distro, of course. I've done it under arch, which is the most cut down of Linux dsitros"
You've done that on a Raspberry pi? Just installed Apache and iredmail?
Quick google says there were Raspberry pi specific patches for iredmail committed to get it to work under Raspbian;
http://www.iredmail.org/forum/topic5305-iredmail-support-iredmail-on-a-raspberry-pi.html
How did you get equivalent patches working under arch?
iRedmail installer says it needs 1 gig of memory to run,
http://www.iredmail.org/install_iredmail_on_debian.html#requirements
so you'll need to disable ClamAV, SpamAssasin, Amavisd etc after install, but presumably you did that too?
http://iredmail.org/wiki/index.php?title=IRedMail/FAQ/Completely.Disable.Amavisd.ClamAV.SpamAssassin
Any other tips/steps to getting this going in under five minutes?
"Just download any linux distro, follow the setup for its apache component and download iredmail, a mix of other open sources apps that installs a functioning mail server in 5 minutes."
Congratulations, you've now just lost 99% of the population.
This is easy stuff for anyone like us, but it is not for the majority of people (even tech-aware people) due to knowledge, time or financial constraints. You have to think of things like an end user would see them. Why do you think Ubuntu has become so popular among people who are new to Linux?
Congratulations, you've now just lost 99% of the population.
As opposed to the what, .03% of people who own a Raspberry Pi? Lets face it, if you're doing dev work for the Pi then you weren't aiming for the mainstream anyway.
This is easy stuff for anyone like us, but it is not for the majority of people (even tech-aware people) due to knowledge, time or financial constraints. You have to think of things like an end user would see them.
So what's the scenario for this? Average user Bob sees an ad online for the Pi, buys one (by accident perhaps?), and then one day suddenly decides he's going to run a web and email server on it but won't/can't be bothered to work out for himself how to do this so ArkOS to the rescue?
I thought that the point of the Pi was to encourage people to get in there and program, fiddle, and mess with the thing in order to computers. This project seems to be the opposite of that.
I've done this on my Raspberry Pi, using Slackware, and the problem isn't the technical part of setting it up. The problem is the fact that you're just some guy on the internet with an IP address. When you try to send mail anywhere it gets returned to you because they think you're a spambot.
Linux depends way too much on a human to get it working properly. That's fine if managing computer systems is your hobby, but for almost everybody it is not.
Secondly, why would any non-techie want their own mail server? What's wrong with the one your ISP is hosting?
You clearly didn't read the bit in the article about not wanting to trust "teh cloud" or big faceless corporations with your personal stuff. I think a home project like this is awesome for those wanting to learn more (which is, after all, the sort of person the Pi was aimed at), but I don't think we should be funding this numpty for a year as per all the other comments. Anyone willing to do it themselves doesn't need him, and anybody not willing to should be renting from an ISP or using Gmail/Wordpress et al in the first place.