(2 replies, posted in NetBSD)

Here you find the list of supported hardware for NetBSD: http://www.netbsd.org/support/hardware/


(2 replies, posted in NetBSD)

kqueue is a kernel service for letting applications know when files or directories change. Using it will usually yield slightly better performance (since you don't need to poll files/directories for changes), but you might not notice it, and cross-platform code like dovecot works without it too.

As you noticed, ssl support you can enable through th eport's options. For sieve support, install the dovecot-sieve port in addition to dovecot.


(2 replies, posted in FreeBSD)

First, if you use tun for your network device, you should use openvpn_if="tun". That should create the tun interface on startup. As for automatically submitting username and password, I don't know at the moment; I use password-less certificates with OpenVPN. Maybe someone else can supply that answer.


(7 replies, posted in Off-Topic)


Maybe accidentally. We're seeing a steady influx of spammers; some of them are the sneaky sort who post properly at first, then add links to their signature after a few weeks. We just delete their accounts and postings; if one of those was a thread starter, the responses from other people are unfortunately gone too.


(5 replies, posted in NetBSD)

Yes, modern software has a lot of dependencies, and it's a lot of work to install something like samba manually. So... the way to do it is to use a pkgsrc on NetBSD. This chapter explains how to use binary packages: http://www.netbsd.org/docs/pkgsrc/using.html#using-pkg
Once you've done that, pkg_add samba should do the trick.


(5 replies, posted in NetBSD)

What is the exact command line you used?


(3 replies, posted in NetBSD)

Hi and welcome, Megan. I hope you'll enjoy tinkering with BSD.


(27 replies, posted in Feedback)

There, spam links deleted, problem solved for now smile


(1 replies, posted in Off-Topic)

Hm, interesting setup you have there. So your firewall ends up aliasing all public IPs for all the servers you have? In that case, I'm guessing it doesn't really matter whether they route it for you or not.

Typically, when you do the routing yourself, you have a router that connects your network to the hoster's network. If you let them do the routing, you can connect all your machines directly to the hoster's network. At least that's how I'm interpreting what you wrote there wink


(4 replies, posted in Off-Topic)

Ah, I see. I can't really say that I have much (well, any) experience on that scale. I'm thinking that it can be done with mail servers, it's just a matter of installing enough hardware. However, the SMTP protocol (and the handling of mails by postfix internally) supports many more features that you're likely going to need for intra-site messaging (even between different instances of your service), which slows down processing (read: you'll spend more on hardware) and increases maintenance requirements and failure risk.

Then again, if you implement something of your own, you might end up with crappy software that is unreliable and limiting, or you might spend more on development than postfix's complexity would otherwise cost you.

One thing you might consider however is thinking of messages and their transport as separate. Define a message format that is suitable for eventual later transmission via SMTP without too much massaging, but as long as your installation is still small and limited to a single database instance, make the transport of messages as simple as you can (e.g., just write the message directly into the receiver's inbox upon creation/


(4 replies, posted in Off-Topic)

If the messages are only sent within your application, that seems like overkill; you could just directly create messages in the receivers inbox instead. If sending emails is actually what you want, then yes, using postfix like that would work nicely.

Give us more details about your architecture, and we might provide more detailed comments.


(3 replies, posted in NetBSD)

Try this:

ifconfig sip0 link 00:11:22:33:44:55 active

where sip0 is the name of your network interface.


(13 replies, posted in OpenBSD)

There is this: http://www.openbsd.org/i386.html


(13 replies, posted in OpenBSD)

You're seeing a kernel panic, one of the more fatal of system errors. Frankly, if OpenBSD gives you this error on a fresh install, without you breaking something manually, it might be quite difficult to get the OS onto your laptop, and I'd recommend switching to one of the other BSDs instead.


(2 replies, posted in FreeBSD)

Situations like the one you describe may arise when the kernel and the userland (specifically, top and the libraries it uses) are out of sync. This seems unlikely though, if you've only upgraded apache and php, and it worked before.

So... does uptime(1) work? Do the numbers stay at 0 even if you generate a high load (for example using: while true; do done at the shell)?


(2 replies, posted in Off-Topic)

Not from my point of view, no. You can always try either irc1.taucher.net or irc2.taucher.net individually.


(4 replies, posted in FreeBSD)

Duly noted.

Why don't you just delete the whole user account. It's not like it'll be used for anything productive.


(4 replies, posted in System Administration)

It might be slightly overkill, but I'll mention it anyway: Nagios. It monitors all sorts of stuff for you by providing a number of plugins (disk usage is one of the default ones, most others check the availability of networking services; also it's very easy to write plugins yourself, they're just shell scripts basically that have to return a status value), and can be configured to do all sorts of things when some set limit is crossed, including of course, sending an email.


(2 replies, posted in OpenBSD)

No, but if you run tcpdump(1) while it dials in, you will see the dns request package, with its source port. If you then quickly run lsof(8), you can see which process has opened that port. Or maybe the tcpdump output on its own is enough if you enable full packet parsing; you should see which DNS address is being requested, that might give you some hints too.


(2 replies, posted in Off-Topic)



(9 replies, posted in Off-Topic)

Many new networking card have an automatic crossover detection. This allows for two PCs to be directly connected without having to resort to a special crossover cable, and it also allows them to connect to the uplink port on switches. The same goes for many new switches, they have uplink port detection.

So the answer is, traditionally, you'd use the uplink port to on a switch to connect to another switch. If you have modern hardware though, you can probably treat it as just another port.

Congratulations to you too. I think August 22nd is a great date too, since it happens to be my birthday wink

/me presents WIntellect with a