Episode 18, Listener Questions #1

PodCast Feed

Links:

Hardware Clarifications
Current costs of running a server at home
One Admin Per Project
Not a podcast idea but maybe an ongoing show segment

Listerner Questions 1

Thud: Since it's been a while, we're going to respond to some of your questions in the forms, and emails, etc. Let's start with the T2000 and here's the audio question.

"Hi, this message is for Seg and I'm going to pre-apologize for the sound quality delivery. This is circling down the highway using a laptop with a built-in microscope, so I'm not sure the audio could get any worse. If you refer to the written message, the question is about your feelings and opinions on the T2000 from the Sun that you'd expressed last week. You seem rather underwhelmed that one of you might explore why. Thanks."
Thud: OK, so the written question basically goes into a little bit more detail. This is a listener that is mostly working at a Linux shop and for some reason, somebody, somewhere in his shop decided Sun T2000 would be a great piece of hardware to run. Seg, what was your big problem with the T2000s?
Seg: Well, when I was young there was a T2000 that killed my entire family and I swore revenge; so I really don't like them. No. At my office, I'm a bit of a jumpstart admin and in-charge of a lot of old things that have to do with Spark and X-36 building. We got our first batch of T2000s about mid last year and it's been a nightmare since then.

From my experience, the T2000 were put out way before their time. In fact, if you've bought a T2000 back then, you couldn't use a RAID at all, because they sent them out with a broken controller card and the RAID was not patched properly. You just could not use it.

After you got the server and re-installed the Sun operating system and patched the hardware RAID firmware, which is an LSI RAID card. Then you could actually use the RAID. Ah! But here's the trick, if you wanted to actually use a RAID, you then had to wipe the drives and start over again.

Again, we found out this out the hard way. We, as a matter of principle or a matter of company procedure, we rebuild every box that comes to us.

If we get a Sun box, and all the Sun boxes that we get have Solaris 10 pre-installed. We install Solaris 10, but we have to do our own build of Solaris 10, which is a bit leaner. It's lacking some of the bells and whistles that Sun likes to throw into their hardware, and it's a bit more secure. It is locked down. If for nothing else, because of the fact that we built it, we can actually vouch for what's on the box.

So, we would get these servers in and we would put them through the jumpstart process, which is an automatic build process, and when they come out, I say, "OK. These are great. Let's mirror the root drives." And damn, the RAID would just die on us. We had to actually power down the server to get it to respond again.

So a little bit of investigation -- oh, Sun just sold us some broken servers. OK, well, let's go get a patch. Let's patch the RAID, and again we want to mirror the zero and one drives. So we patched the RAID, we back up and try to use the RAID utility that we are all so familiar with, called RAID Control.

RAID Control gives us the middle finger. It says, "Nope. You're not building a RAID because I'm using the zero drive." And I said, "Well, I know you're using the zero drive; it's a mirror. Just copy what you've got over to drive one." Again, RAID Control gives us the middle finger, says, "You're not doing it." It's OK. A little bit more investigation later, we find out, "Hey look, you can really only do this from the OK Prompt, which is ironic because any Solaris admin worth his salt would say that, when you're at the OK prompt, usually things are not OK.

But anyway, you have to build the mirror from the OK prompt, and that takes many hours. There were 72 gig hard drives in there; SAS hard drives in there, and it took a few hours -- maybe six or seven hours. I don't remember exactly. By doing this, you effectively destroy the data that is on the drives, because it changes the size of the drive at the hardware level, or at what the OS thinks the hardware level is.

So, then you have to repartition the drives, and then rebuild the OS on to it, and so it took a process. It took a process which -- literally our jumpstart process takes an hour and 45 minutes. It takes 50 minutes to actually install the software, and then all of the patching and lock downs -- everything we do after that takes another 45 to 50 minutes. So it took a two hour process and dragged it out for days.

Our first batch was like 62,000s and it was just a freaking nightmare. Things got worse later on, there was a card. As I recall, one of the actual parts of the server, I can't remember the damn name of the card now, but it was a card, an internal card, wasn't like a PCI card or PCI-Express card or anything like that. They actually had to send Sun technicians out there or subcontractors to swap these cards out, which was just insane. Of course, our customers were like, "What the heck? What are we buying? What are you guys doing?" It was just crazy.

Thankfully in that particular environment, not a lot of money was lost. It was pre-production; it was just about to go to production. Again, the things go on from there. The RAID was a big deal. But things go on from there. I'm familiar with ALOM and LOM. ALOM is Advanced Lights Out Management, LOM is Lights Out Management.

The function of those is to give you some kind of control over the very base hardware level access of the machine, when the OS is not running, not even the OK Prompt is running. If you take a Sun Fire V240, it's got ALOM on it; you can plug in the power cords to it, not boot it up at all, just plug in the power cords and plug in like a laptop to the console switch, which is at the console port, and you can access the ALOM.

Server is completely off, but it's got power. So you can do things like, you can tell it to power off, power on. You can set some of the boot variables and do so many things.

To get into that though, and to make any kind of powerful commands, such as you can send a break -- power on and power off; you have to have a password. You have to set up an admin account. If you lose your password, but you can still boot with the OS; on a 240 it's no big deal. You boot into the Sun OS and there is a program off of the Platform tree, and again I forgot gosh-darned name of it, but you can reset the password for any user on ALOM from there.

On the T2000 however, you cannot. You cannot do that. The only way to do it is to actually boot it down, and you have to reset the LOM, and just wipe out all the variables and go back to factory defaults, and then set up a new password. It's a gigantic pain in the ass.

And then we moved on to OBP Patching. Again on a 240 -- one of my favorite boxes is a 240 -- if you need to update the OBP Firmware, you want to update Post; it's real easy. You download this file, you copy it to the root directory, you make sure it's chmod 755, boot down to the OK Prompt, and then you type, "boot disk", and then "/", and the name of the file. And it boots off with the disk and uses the file as a parameter. Then it will boot into that file and say, "Oh, you want to upgrade your boot prompt?" and you say, "Yes." It takes about four or five minutes tops, and does what it needs to do. I've never ever had one of these things fail. You do a reboot -- bam -- everything is done.

On a T2000, that is not the case. You need an FTP server to update the OBD and the boot prompt. It's a nightmare. We don't set up FTP servers in our office. If we can help it, we get away from FTP as much as possible. SEP is the wave of the future. Encrypted -- everything over SSH is the way to go. But if we intend on doing the T2000, we have to set up an FTP server and it's just a nightmare.

I haven't even done it because I don't want to do it. I look at it like: This is just ridiculous. If a patch comes out where it's critical that it has to be done, fine, I'll set up an FTP server on a jump start server and go forward from there. Beyond that, no, I'm not going to do it at all. It's so not worth the time.

[sigh] Now if you can get past the patching, the raid and the faulty hardware, fine, go with the T2000. Your jump start admin's going to hate you for it, but once it's actually built and running, you're going to love it. It's a lot faster than T240. It's got the four internal drives -- they're SAS drives, PCIX across the board. Real fast bus speed and it's a really, really fast machine. That's really what I can say about it. It's a cool, fast machine.

At one of our racks in my office, we've got about 15 or so on one rack. If you stand behind there, you can cook some bread. It's not as bad as some of the other servers, and the servers themselves do stay pretty cool.

The root of the whole thing is that some push these machines got out too early. They got out there without the needed testing. I'm sure they were tested to a certain degree, but they needed a lot more testing before they hit the market. They've been out there for a while now, a little over a year. I learned about them a little over a year ago.

They've got some of the quirks worked out. In fact, if you buy a new one now, they will come with a RAID that's got the proper firmware. That was what really kicked me in the nuts back when it happened because they sent the server out pre-built and they had the firmware available on the Sun website but the two weren't connected. They were sending them out without that patch. You had to go and get that patch which is crazy. I can kind of understand they sent it out when there was no patch but they were still selling them and shipping them out without the patch.

Nowadays you get them with the replace card, with the right firmware and they will actually work. Whenever possible, I try not to do it. If for no other reason, the alum password is a big deal, and the RAID. On the 240, if you've got two drives, and you're using one of them, you're using drive zero and you just want to say, "I'm going to pop in a second drive and make it a mirror," then you can. You use RAID control, its hardware level, then you have a hardware mirror and it's fantastic.

But you can't do that with a T2000. If you don't have a RAID and you want a RAID, you have to rebuild the entire box. There's no way around that. In fact, a couple of our T2000s went through the process such that even though they paid for the LSI RAID card, and they paid through the nose for it, and the server can do hardware RAID, and we recommend a hardware RAID on these, they're not using them. They're using Solaris Volume Energizer as a software RAID. They can't take the damn thing down to do a rebuild, and it would just be way too much of a pain.

So that is my list. That is my entire list of why I hate T2000s, and whenever possible, I won't use them.
Thud: Yeah, and I would like to point out that it's not just a Sun issue. There's plenty of manufacturers that release hardware too soon. A good example is, right now, a problem we constantly have with new Intel servers in my data center is they come with RAID cards that, when you first boot up the system, the RAID BIOS comes up and it tells you that it's a beta version and that it should not be used in production. And this has been going on for six months.

We ended up having to patch the firmware on the RAID card, which means, because it's now incompatible with the BIOS on the motherboard, we have to patch that as well. And by the time we get done, we've spent the first two hours of work on the server just patching all the hardware. I don't know why hardware manufacturing is going this way.

Another good example is I bought a network card--a network card of all things--and the very first paper in the box, before I even get to the network card to take it out of the box, says, "You need to download the firmware from our website to make this functional." You know, take the extra 10 seconds to put the right version on there, and then ship it to me.
Seg: You got lucky. You got warning on that. When I ran into the T2000s problem, there was no warning at all. I had to actually hit the brick wall a couple times. It's not working, it broke, and I had to go look it up. But yeah, it's horrible when these things happened.
Thud: Yeah. It's caused me to have a general rule of: when new hardware comes out, let it be new for six months, and then worry about trying to bring it into your systems. Just invariably, there's going to be problems with it, and let everybody else be the guinea pig.
Seg: [laughs] There are always going to be plenty of guinea pigs, too. Yeah, definitely. One of the things that are drawing people to the T2000s, especially the customer at my data center that uses them religiously, is the cost. Apparently, there's some kind of like, "We'll give you 10 for free, and five years down the road, we'll charge you $10."

There's some kind of amazing pricing deal that's going on with these things that people are just gobbling them up. They're like, "Oh, this great server for such a low price!" Well, there's a reason for that: it's because they suck.
Thud: [laughs]
Seg: But the managers, the people who write the checks, don't see that, and it's hard for them to even understand why these RAID problems and everything would be an issue down the road. They're like, "Well, it doesn't matter. It's so cheap."

And one of our customers uses Sun x86 products. I really wish they wouldn't. And again, it's a cost issue, like, "Well, we can get this amount of computing power for this amount of money, or this for so much more." And I'm like, "Look, you're paying for the quality. If you go to Sun, you're going to pay through the nose, but you need to go with the SPARC hardware stuff. If you're going to go x86, don't even bother. If you want cheaper hardware, then just run Linux. Just run it on Intel hardware."

Sun x86 is all crap, and especially the new stuff. There's an X2200 out right now that's just a total piece of crap. The Solaris engineer at my office got the privilege and the fun of working with that thing, and all the documentation was wrong for setting it up on the console switch. I don't know what he ever did to ever get it to work on the console switch. [laughs]

But all I know, I wasn't around him a lot when he was doing this, it was like five weeks where this guy was just livid every single day, trying to get this thing to work, and finally, finally got the damn thing to build and boot up. But the company that bought them were like, "Hey, they're real cheap. Let's just go with that and let our data center worry about how to build them." Well, thanks, man. You're killing us here. Again, I wouldn't go for them.
Thud: Yeah, I don't know why people even think about Sun and low cost. It's not often that I recommend Solaris in a production environment. It's normally only when there's Oracle going to be there for a database server. And the minute I think Sun, Solaris...

The minute I think Sun, Solaris, Spark, the cheap server I'm going to buy is going to be $100,000, because if you buy anything lower than that, you're probably going to end up with crap. The software and the hardware are designed to be on those huge, big systems. That's where Solaris and Sun hardware shines.
Seg: Yeah, yeah. If you go and buy V240, which has been out for a long time -- but it's not end of life yet -- you're going to pay something like two to three times as much as if you buy a T2000. But I'd still, for my money, if I had the money to spend I'd go with the 240 every time. The old Spark stuff that's been out for a while, the Sunfire series, the V240, 440, 480, 880, 890s, those are all fantastic pieces of hardware.

If I really had the money, I'd go for an 890. This gigantic machine that's like a quarter of a rack high, but it's stable and the damn thing works really, really well. We've never had a serious issue, or even a minor hardware issue with an 880 or an 890. If you go back to the old stuff -- the Enterprise, the 450s, the old 280Rs, although even the 280Rs are older than dirt, but people still use them because if you can get the right parts together they do work pretty well. So just wait, just wait.
Thud: [laughs] All right, I think we've covered that pretty good.
Seg: Yeah.
Thud: Let's move on to some of the forum posts we've got recently. The next two are about home servers. The first one is by Simace2 and he has a couple questions about the hardware clarifications for a home server. The first one is: What physical setup do you recommend for a home server: a box in the corner that you log into from your desktop PC, or do you have a rack with keyboards and a small monitor. What do you have set up at home, Seg?
Seg: This is what I did; the one thing I can really call a server in my office. I went to Newegg and I wondered how cheap of a desktop can I build at Newegg? Not counting the hard drives which I already had, I spent $200 to $300. I built the cheapest machine. I got an Intel board -- a P4 was the cheapest thing they had there -- a P4 processor. I got a couple of six gig Ethernet cards and the cheapest flimsiest case I could find and it's fine. I run Free DBSB on it. It works perfectly, no problems at all.

The caveat to that though is I don't do a lot on it. I use it as a firewall router, and that's about it. I run SSH through it so I can get to it from my office and that's all it does. I wouldn't serve a production website off of it or a production anything off of it. So, the cheapest thing is going to work for me.

But that's the question you have to ask yourself: What are you going to be doing with it? If you want to run a home server, it can be very cheap. But if you're trying to do some big websites, you want to run a game server or anything like that, then you're going to need a bit more firepower at your disposal and that's going to run up a little bit.

Ultimately though, servers at home are not really that good of an idea. In most cases, servers are meant to be serving production applications, production website, and those things belong in data centers, they belong in dedicated locations.

I had a dedicated solution with a company called Layer Tech for a while there, and it cost me about 80 or 90 bucks a month, but it was completely my own system. You can go much cheaper than that if you want virtual private servers, you can get them in the four to five dollar range. You just got to shop around a little bit. It just all depends on what you can afford.

If you just want to host a small website so you can share files with your friends or something. Yeah, two or three hundred dollar machine off a Newegg and you're all set. That's all you really need.
Thud: Yeah, I actually, over the last year I've actually converted. I started off with a wire rack in a closet in my office. It's got my Internet connection, my switches, my file server. I probably had six or seven machines in there. I got them from old work stations that I had upgraded to something new and turned it into a file server. I got a couple of good deals on eBay, or even like Newegg refurb or something like that. Just so that I would have a lab that I could play around with.

I still have some of those machines, but for the most part I've switched everything to VMWare. If I want to set up a lab to see how FreeBSD works as a file server for Windows, I just set up a FreeBSD VMWare machine, and a Windows VM, and play around with it that way. It's a lot easier to manage. It doesn't mean I have to have a little beefier workstation, but it makes it a lot simpler to set up.

For me the two most important things were, one my office was the hottest room in the house, because of all the heat put off by these machines, and I could stand to use a little less electric bill every month. [laughs] So by doing VMWare it solved a couple of my problems.

But for the most part I wouldn't recommend trying to run, as you were saying any kind of production anything out of the house. It's one thing to have a lab at home, but to have mail servers that you really expect to receive mail through, maybe give an account out to a friend or whatever, put your website on. For the most part, especially in the US, if you have some kind of broadband Internet access. The company that sold it to you doesn't want you to do that.

There's also a downtime issue. If you go with a server in a data center, even some cheap dedicated server, or virtual site or whatever, your going to have much better up time with that. And besides, if you're paying somebody a hundred bucks a month to keep your server up, at least when it's down you have somebody to yell at besides yourself.
Seg: [laughs] Yeah, home servers are great for labs, like you were using them for. If you want to get into the business, you want to learn more about it, then you absolutely set up a home server, but Thud's right, don't expect it to be production level. Don't put your mail server on that and expect your mail to work all the time.
Thud: All right Seg, what are some good places to buy equipment, motherboards, RAM, whatever?
Seg: I mentioned a minute ago, New Egg and newegg.com is the only place that I go for hardware equipment. I've tried Tiger Direct. Absolutely cannot stand them. I know that I'm one side of the fence here. I know other people who speak the same way in the opposite direction about New Egg and Tiger Direct. My boss for one swears up and down that Tiger Direct is the best place to go ever, and that he hates New Egg, and I don't get it.

I've been using New Egg for a few years now, wouldn't shop anywhere else. I've hit up eBay a couple of times for some really hard to find parts, but those have all been Sparc based equipment. I needed a, crap I don't remember the name of it now. I needed this old 2000 year old Sparc part for a fiber array -- a storage array of fiber. A G-Bit, that's what I needed, and I found it on eBay.

[laughs]

Because I couldn't find it on Newegg. I might try Overstock.Com for something, but I limit that to, if I need a stack of writable DVDs or CD-Rs, I might check out OverStock.Com for a good price. Outside of that, the only site I can recommend is Newegg.
Thud: I actually use Newegg a lot and the other one I use, besides eBay, is ZipZoomFly, just for comparison because usually between Newegg and ZipZoomFly, I can get a good deal on whatever part I am looking for. Usually one is quite a bit higher than the other. But I can't ever go there and build a whole system at Newegg and get the best price.

The last time I did it, I got everything that I needed at a good deal at Newegg, but the hard drives were really a good deal at ZipZoomFly. So I got the hard drives from there, but especially for a home lab system or a multiple systems in a home lab, a lot of times eBay is really hard to beat because you can get...you can use almost any hardware you want for a desk server, it doesn't have to be a server, it can be a desktop, and there's always someone out there who just upgraded their system and wants to get rid of their desktop for a couple hundred bucks.
Seg: One actual note, one of the things I love about Newegg is the review section. It's a user submitted reviews on all these products. I went and I bought an Acer system, because I wanted an uber-cheap desktop that had to do very little work and I didn't actually want the hassle of building my own system. The downside of Newegg, as I found out, is the only prebuilt desktops that they offer are HP and Acer.

I'll never buy another HP machine because---I don't care how good they are now-a-days---I got burned by them a long time ago. Never, ever, ever buy HP ever. I'll never support them; I don't want anything to do with any kind of Hewlett-Packard hardware. So, I bought an Acer machine and I got kicked in the head on that one too. This Acer that I got it runs, but it's a piece of crap. Just recently, I unplugged the headphones from the headphone jack. It rebooted the machine.

[laughs]

Yeah, it was a nightmare, just trust me, dealing with the whole thing, it was just a nightmare, I'll never buy them again. If you want a whole system, go with eBay. But if you want to build your own, then go with Newegg.
Thud: Yeah, I've actually got burned a couple of times too. That's why, if there's a possibility I'm going to get burned, $200 is my limit.
Seg: Gotcha, yeah. Absolutely.
Thud: If I spend a $1,000 and get burned, I'm going to feel horrible. But if I spend $200, I can find something to make it work. All right, our next forum post is, I just invented a new word there, forum post, is about the current cost of running a server from home. And we've kind of already touched on this.

I don't recommend running a production server at home, mostly because in the market right now, you can get dedicated servers for well under $100 a month. They are not going to be...they are actually going to end up being cheap desktops that are called servers by the companies that provide them. For most people, that's more than enough. And it's going to be in a data center, hopefully with decent cooling, ups, a generator maybe, somebody to run the Internet connectivity because the one thing that I've had some friends run into is you get...

If you have an ISP, whether it's dialup or broadband or whatever, they are not going to like you running the servers unless you are willing to pay for business services and some of them are ridiculous. I have Fios in my house, and it's fiber optic from Verizon, it's great bandwidth, it's amazing. But I would have to pay four times what I'm paying right now to get the bandwidth I have right now but get the business class line which you allow me to run a few servers. It won't let me run whatever I want, it's limited to mail and web. It's just not worth it. For $80 a month, I can get a dedicated server almost anywhere in the country and do whatever I want to do with it. What's your advice, Seg?
Seg: I completely agree with you. When I first got into the idea of running something from home, running real services from home, I was working for this call center, it was an ISP. And I probably, in theory, if it had been a good company, could have gotten a discount but I looked at their base prices for running a web server off their network and it was just insane.

I agree. If I am going to run my own stuff, I'm going to want to put my server in a datacenter. First off, it helps keep people like me in business, but it's just so much more of a wise choice. You are going to have the support you need, you are going to have the uptime, you are going to have the person to yell at when things go wrong. Although you shouldn't yell at people, but yeah, it's a good way to go.
Thud: all right, now we are going to move on to her detailed response from one of our listeners to the one admin, one server episode. I'm trying to find that guy's name and can't find it.
Seg: Lokes? Yes, Lokes.
Thud: Lokes, yes. OK.
Seg: It says, "Guys, I want to really agree with this cast, but just can't. I left my opinions on my blog," and he put a link to his blog. This was the "one admin, one server" pseudo-rant I did. One of our previous episodes, episode 16, the idea of that cast was to portray the ideological viewpoint that for every system that you run, for every sort of application you run on a system, be it a web server or a mail server, whatever, you should have at least one person for that. If you go in that direction and you hire people to work on projects and focus on things then you are going to have the support that I think is required for each user's applications.

Lokes says that he wants to agree with this, but he can't because he lives in the real world. He says that in the real-world, I guess he's like a freelance sys admin. We don't have the time or the resources to do that kind of thing. And I try to understand, and I tried to portray that understanding in the podcast. I understand that, saying, "OK we run this one website for this one server, we have this one guy."

And then for the mail server, we are going to have this one guy. And for this other web server, we have this guy running it, and so on. Before long, it's over 300 for one company, and it seems like a lot of people for such a small thing. I understand that in real life you really can't do that. There aren't enough sys-admins in the world to fill those kinds of quotas. Certainly in my life I have to touch hundreds of servers every single day, and tons of applications, and I lose track, but I get it done.

But I want there to be a better understanding that I can juggle 300 servers every day and just keep them up and barely working, but if you want me to do a really good job, you've got to let me focus on one (or two tops) of these projects, because there's so much going on on the Internet that I need to be up to date. Every minute of every day there are new bugs and new security problems. All that stuff's constantly changing, constantly growing, and I need to keep things constantly patched and constantly changing. And if I'm juggling 300 servers, it's really impossible to do that.

That's the direction I wanted to go. I wanted to say that I know it's not possible to do it, but I want people to try for it. I want us to understand that that need exists and we should at least go in that direction. We should admit that it's not OK to just have one guy for five servers handling db-mail and web applications and all the sub-applications that support the website. It's not OK for a web developer to be a sys-admin if he's not qualified for it, and vice versa. You need to have separation. You need to have multiple people on these projects helping out in every facet possible.

Lokes made a good blog post. I read it a couple of times and I did enjoy it. Thank you very much for posting that.
Thud: I have to agree with that. In the real world, you're always going to end up with one guy who's the top admin for that project, but you definitely have to have people that can fill in behind him so that if he were to get hit by a bus or have a meteor land on him, at least your systems can be up and running.

I think if more companies realized that it's a bad idea to have one admin in charge of five or six different things with no kind of backup, there'd be a lot more admin jobs out there but it would also be a lot less stressful job. I can't tell you the number of times I think, "I need to take a vacation, but I can't, because there are things that won't get done because I'm the only one to do them."
Seg: Unfortunately, there's a vacuum out there. There are not enough good admins. There are apparently plenty of people who want to be admins and who are not good at it. The people who love it, who really live for this kind of work like you and I do, they're very rare. And they've all got jobs being admins.

I really wish that more people would... I want to say I wish there were more good admins, but I also believe it's something that you have to have a passion for, and I don't think you can learn a passion. It's more like something that's in your blood, really, as poetic as that might sound. I'm doing the job I really love to do. Some people will climb mountains, some people bake cakes, some people paint. I want to be an admin. I've found what I really love to do, and now that I've done that it's just a matter of trivial points. I really wish more people had that kind of love and passion for this kind of work. I believe that's what makes a good admin. At least that's what my co-workers think.
Thud: I have to agree, and I really wish people had more appreciation of the work that we do. I know people that call themselves sys-admins that are making twice what I do but have a third or a quarter of the work load that I do. I'm in the same boat. I really, really love what I do. When it comes down to it, as long as Visa and MasterCard aren't calling me every five minutes, I'm happy. You've got to pay me enough to cover that, and I'm going to continue to work as hard as I possibly can, because I really enjoy what I'm doing. It's not about the money.
Seg: I've worked around people who are just in it for the money, and that's a tough situation. In some cases, in some companies, it's good to have people there just to help out; to be a warm body in the chair, as some say. But in other cases, if they don't love the job, they're not going to be the full support that's needed at all times.
Thud: Not only that, but they're not going to be happy employees. It's not just sys-admins. I think it's very important on a personal level that whatever you do to make your money is something that you love. Doing that is going to make you a much happier person. I started off my career as a data entry clerk, and within seven or eight minutes I started hating paperwork. [laughs] I'm glad that I found the sys-admin career, and I think that's important on any level. You just have to love what you're doing.

All right. Do we have anything for a moment of sac?
Seg: Not that I know of.
Thud: OK. So instead of a moment of sac we're actually going to discuss another forum post that just came in recently. We want to get your feedback on this. M. Tiernan posted on the forum a suggestion for not really a show but an ongoing segment on this show called "Best Practices." He says every podcast, just take one little thing and discuss some ways of doing it. Anything from sweeping the floor to naming the servers.

Generally I use a broom for sweeping the floor. If I used a vacuum it would be vacuuming the floor, and that's not sweeping. If the broom is wet then it's mopping. [laughter]

We want to get your feedback on the idea of along with having a moment of sec (which, oddly enough, we're missing in this episode), having a "Best Practices" moment as well. So if it's something you want to see us do in future episodes, please post on the forum, give us some ideas, and we'll see what we can roll into future episodes.
Seg: Also, we both want to say we apologize for such an extended absence. We have been fighting ninjas and PERL for weeks now and it's just been hectic. We had to pull some miracles just to make this show happen. But we're still here. We're still plugging away, and we still want to continue to make good shows. So thank you for your patience. Thank for sticking with us.
Thud: Seg wants to apologize, but I don't. I'm not going to apologize for not sleeping for two months. [laughs]
Seg: [laughs] I was trying to be nice. Real admins know what we're going through.
Thud: Like we've said before on past episodes, we do this for a living and we are Internet firefighters from time to time. So when we have five minutes to record a podcast, we generally try to sleep or eat.

But we do actually have some that are in the process of being edited and put out. This will be the next episode, and we have two that we actually recorded earlier that will be out in the next couple of weeks. We felt it was a good idea to go ahead and record this one, since it has been a long time and there's this backlog of things to respond to. So thanks for listening.
Seg: [laughs] Thank you for your support. And I guess that's it.
Thud: Somebody send me an iPhone. [laughs]
Seg: [laughs] "Make checks payable to..."