Feedback on the Linux config guide.

For reference: This guide was made in 2013 and then pulled in 16 months. I made this at a time this was a common problem with my guides due to severe quality issues. If you were expecting this to be a preserved rehab, let me disappoint you now. It isn't. I tried, but it wasn't worth the trouble.

So, this is what the 8 month delay got me. While the first 2-3 months were wasted (more on this later in the post), I made a call on how to get rid of that delay. It turns out that call to get rid of it was worth it. If I had opted to go with the original plans, this would have taken years to make. It was a effort that proved to be futile, at best. In regards to the text part, the last 6 months were spent making it work as a new guide. This basically killed a lot of my problems of making it work with the original content.

NOTE: While this is a public release, I am only keeping it public for 2 weeks so I can see how much I missed. I knocked a lot of the issues out, but I probably missed something seeing as I tried to keep it intact with modernization with years of newer hardware, which screwed it all up. Right now, I have to find a system but I know what I want. Until I find one I will be 45% behind on pictures and more or less 5-10% on content. I am WELL AWARE the guide is a work in progress. It will be made private again on the 25th.

• All of the steps were rewritten from scratch, but not replaced. I still have the 2013 edits left, as I used the original steps.
• I have decided hardware minimums had to be put in. You can read more about it in the specs in the guide if you want to see what I decided on.
• I chose to leave Santa Rosa and Conroe behind. There are 2 reasons behind this. I did this because of line limitations vs what made sense to keep and get rid of. This is why I let SR/Conroe go and put Peryn/Wolfdale in place. I feel like SR/Conroe are nearly at the end of their life, if not very close. I do not expect the average reader to know that these chips are old enough their time is nearly over for even basic tasks. To me, anything older then Peryn or Wolfdale is old enough I feel it's time to move on. I also wanted something Core 2 related in the guide, but I didn't want it to be so old that the performance is no longer acceptable.
• Merom was dropped, even before Santa Rosa. It has been obsolete for years. Being a gen one mobile C2D it has serious performance issues that have been showing for years. I feel like the performance gap is already bad enough as it is.
• The age limit for hardware is set at 2007 for desktops and 2008 for laptops, up until 2012. I tried to go all out and update the guide with many, many years of hardware and it did not work. I decided to put a 8/9 year old cap on the hardware to make it easier to get done, and for readers to follow. Generally speaking, a system is considered obsolete after 10 years. This always isn't the rule as there can be exceptions but for most systems 10 years is the point it is considered obsolete. I didn't want to go to 10 but I also didn't want to make it too early.

• I have added steps for wireless. This was something I never added because of the complexity at the time. I just didn't know enough to incorporate it at the time. Now that I do, I had a bit more confidence putting it in. I also wanted to wait until I got the core guide done before adding it to keep the progress I made at a normal pace.
• I added a step for new hard drives. This is also a new addition because I have learned how bad some of them really are. I have seen a 7200.11 give the BSY error and crash the ATA Security set, a .12 have SMART issues and a .14 in it's natural habitat, which is needing replacement. On top of these 3 drives I have seen others start to have problems after 3 years in some cases. However, I am not mandating a new drive. I understand some people want to reuse it and see, so I added testing suggestions to make sure it's healthy, or just said it probably should be replaced so you don't have a age related issue. Very often, the health is mediocre to poor and the remaining life is 1-2 years. My opinion is they should be replaced but I know some people want to reuse them. I would rather put testing instructions in to save them the grief of not knowing how to verify the drives for long term stability.
• The SSD step is new. At this point with cheap 256GB and 512GB SSD's they are not super expensive anymore. I do not have a new and used SSD step because of how new they are. Hard drives have been around forever, so I need two steps for that.
• I have depreciated the chipset step, but did not remove it from the guide. I was going to remove it but I felt it was better if I depreciate it but keep it around.
• The Linux variants have all been replaced. I removed the original ones I had there back then in favor of some more targeted ones. NOTE: The board with the chipset picture is the board from the old system. Consider it an Easter egg from the original guide. I kept it from the partout as a CPU holder due to the weak VRM and lack of Pentium D support. I am not mentioning this very much because I want to see how many people recognize the Conolly marker as being from a Dimension 5150. I didn't removed the heatsink and exposed the i945, but because I may need to reuse the board and 10 year old PIII style thermal pads are a nightmare to remove after 10 years I chose not to.

Part 1: Getting a actual system.

The second part of the 8 months was spent finding a system. I got a HP IQ506 in April, but I needed 3 parts. I got it under the guise the parts would be easy to find, but it turns out it was quite the opposite foe one of the 3 parts, while the second one was only semi difficult. On top of that, I needed to find the screws. The screws weren't a big deal. I was able to get the thread numbers with a digital caliper that I need for 4. I could then take that information to a hardware store and get flat top screws. That was not a big deal. I also needed a bunch of other screws but I had a sample I could bring to get that sorted so I was in good shape for the screws I needed 11+ of. One of the good things about this system was HP only uses 3 screw types and they don't use different length screws.

I needed the following parts to get it going initially:

• Stand ($40-40 used; $70-80+ NOS; difficult to find)
• RAM door ($10-20; semi difficult to find)
• Power supply (varies widely; sometimes $20 but can be $50 or more; easy to find)

Due to the cost of the stand, I decided I am better off getting a parts system. If I got a stand alone, I at least had the security of knowing you can buy the RAM door with a bit more ease compared to the stand. In the end, I used a parts system. This was because I gave up finding the stand so I figured a parts system would be easier to find. The condition of the parts system made it a literal parts system. This is the failure list it had:

• Screen (Solid TAB failure lines; this happened on the iGPU LVDS connection as well as the dGPU LVDS connection)
• GPU (HP white screen with colored lines syndrome)
• MXMII GPU heatsink
• Touch controller
• Hard drive (7200.12; SMART data could not be read.)

It was a parts system in the literal sense. A fair amount of parts on it are bad to the Since I only wanted it for the stand and RAM door, this was okay with me.

The only parts that are good are:

• RAM
• Motherboard
• Inverter
• CPU
• CPU cooling fan
• GPU cooling fan
• CPU heatpipe cooler
• Wireless
• TV Tuner
• Speakers
• DVD drive
• Chassis components

Outside of this list, the system was shot beyond recovery.

Part 1.5: Getting a power supply

About a week later, I ended up getting the power supply I needed. This was $20 and I got it from the source I got the dead system from. This was it for initial parts to get it going.

When things started getting bad

Part 2: The system acts up after being repaired

While I did manage to find the parts, it started having backlight problems in May. At that point I decided to look at the lifespan left on the other parts before I decide to repair or consider it a total loss like the parts system. What I found was the condition of the drive was a lie. It was actually used and it had problems 2 years in (2014 manufacture) and the seller hid that by resetting the POH. It also had major SMART problems. Having to replace the drive was going to be another $50-ish on top of what the screen and inverter costs.

What it would cost me to fix it:

• Screen/Inverter ($80-100+; inverter was made of unobtanium. I usually change both to ensure the problem is gone. An inverter is usually $25.)
• Hard drive ($70; WD Black 1TB)
• DVD drive(12.7mm slot load; usually $20-25 on average)

I priced the repairs out and I have decided the machine is a total loss. It's going to be WAY TOO EXPENSIVE to repair this machine. I still use it but a failure will be the end of it's run. I'm more annoyed at the fact it still had problems after the time I put into it more then anything. I didn't expect the Seagate to be reliable in the long run because my experience with Seagate is their drives are crap these days but the fact I took the seller on their word on the drive annoys me the most out of any of the failures it has had. The tip off was that the performance wasn't good on the drive, which usually means the drive has a problem.

Part 3: The future beyond this system

When issues like this happen, I try and look on the bright side as best as I reasonably can. On one hand it sucks I got screwed and the repairs were a waste of time. On the other hand I basically have a strong supply of these parts, which will give me an advantage if I need to repair one of these systems. In a way, I lost but in another I won. The systems are both a loss but I have 2 parts machines to pick from if I need anything for a repair on this series of system later on, with various problems and missing parts. The downside is I need a system again.

Am I going with the IQ506 again? No. I've had 2 losers and do not want to go down this path again.I am going to be picking something a lot newer this time. As much as I'm sure the repairs would be welcomed I feel like the IQ506 is at the stage where repairs are rarely viable. You can fix it, but economics no longer make it worth it unless you have spare parts laying around.

Despite ALL of this, I was able to push forward even without a system even though I intended to use the HP IQ506 initially, but decided to move on from to avoid these problems. I don't believe the third is a charm. After the 2nd has a problem I will walk and pick something else. I'm prepared with the information on what I want so this is going to be easy to achieve.

Once we get past the initial part situation and then dealing with the major failures the second time that did it in for the IQ506, you have the remaining 4 months of the 8 months it took to do it. My enthusiasm about finding a system that was any good was burnt at the time due to all of the shenanigans, but at the same time I didn't feel like dropping out entirely because I felt I could make it happen anyway. I decided to refine the guide until my enthusiasm came back so I didn't ditch it but I also had time to think long and hard to avoid further problems like these HP AIO's have given me so far.

I was in a much better position without a system. I was not tied to a age and could do what I wanted, which made this process so much easier. Why not write the guide while imagining the good enough and perfect system at the same time, rather then chain myself to a 9 year old AIO that has a limited upgrade path and has given me problems with 2 examples straight? I can always buy the system when I finish the guide if I don't find a deal or buy it if I see a deal and mention it when I get the text done.

Why 2008 for laptops and AIO's?

Multiple reasons.

• This was when the first Peryn processors were released. I lined my hardware years up with the processors and left the rest neutral as there are times when some hardware's life is extended beyond the processor.
• I feel more comfortable with 2008, as this is when LED panels started coming in. This is because I am finding cheap CCFL panels are starting to fail in mass in recent times. The good ones aren't as problematic yet but the cheaper grade ones are showing their age. I'd rather avoid CCFL as much as possible for that reason alone.
• CCFL repair costs. When a inverter fails in a CCFL based system I like to change both to make sure the problem is gone. I can mix and match parts to a degree, but I don't like to do it beyond 3-6 months at most. I can go longer but I will only do that if I know the history and how it was treated. I usually do not know this information. As such, I change both and get it fixed in a way I can trust in the longer term. Lately, it's been getting very hard to find CCFL parts. Most of the panels are 5+ years old with many, many hours on it and you usually don't get it with the inverter included. On top of this, the prices have been going up in recent years due to the demand and lack of supply. The systems can be fixed, but the market has made them too expensive to repair vs getting a working one or going LED.
• For AIO's, it's panel class. You will end up spending more to repair an AIO then you will on a new one. It's almost never worth trying to repair AIO systems based on that.

Why 2007 for desktops?

• Wolfdale came out in 2007. Again, matching years with first generation releases.
• No panel concerns, seeing as most machines have a seprate LCD. If the LCD dies then you buy a new one and be done with it. It's not a major concern because of that alone.

How I feel about going beyond my specs I decided on

I picked specs that I could feel comfortable with the person building their first Linux system to get them in the right direction, so they don't have to fight performance issues on legacy hardware. I know from experience when old is too old.

If you can figure out how to make it work, then I don't have a problem with someone using a SR laptop or a Conroe desktop or even Merom for that matter. My official position is any hardware that isn't up to par with what I decided on isn't going to be supported.

However, I do encourage people to try and go beyond my first Linux PC specs if they have the skills to make it work. It's just something I will not be supporting and will be treated as a try at your own risk sort of deal. I will let it happen, but I am choosing not to support it officially. I am treating it as an official exception.

Why did it take 8 months?

It took 8 months for many reasons. Let's talk about the 2-3 month delay from the retrofit period first.

More information on the 2-3 month retrofit

I consider the first 2-3 months were wasted trying to make it work on the original content. After that kept causing me trouble for about 1-2 months, I decided to drop it and start with fresh content and treat it as a new guide rather then a legacy and modern hybrid. The problem was it made the guide hard to manage. In the hybrid situation I had 3 year old content that was written so badly it had to be replaced, combined with recent hardware and text that is written much better then the original text was. The accommodations I would have had to make to pull that off would have been time consuming and difficult. I would have to determine what is legacy content and modern content, and split it with more steps because I don't want to use all of my lines up and mix modern and legacy content together as this would be a waste of my 8 lines I have to work with. After all, I need it for the modern hardware too. Once I decided on what to consider legacy vs modern, I'd need to write the new text for the modern hardware and then revise the legacy content, if not just rewrite it altogether. This would have been a hassle and a huge tine consuming thing, which is what settled the decision to get rid of it. On top of all of that, some of it would have to be replaced because I blew it away earlier in the corrections. The only legacy hardware left was processors when I thought about it, so all of it would have needed to be replaced. I even blew away some of the CPU's in the process of doing that.

While it would have been cool to have fresh content and legacy content, it's also more of a hassle then getting rid of it and replacing anything I deleted that can't be interchanged. It also doesn't make sense to do that in a environment where you are trying to make the guide modern. That's why I decided to drop the idea and start over without the legacy content.

With that out of the way, here's the link. I think I have everything dealt with, but maybe I missed something.