Compaq DL360 Multiprocessor Problems

As I blogged before, I’ve been having some fun with some Compaq Proliant DL360 servers we recently acquired. After fixing the Fedora Core 5 install issue on the first server, my second server installed correctly, but upon booting up, it only recognized a single processor. No matter what I tried, it wouldn’t see the second processor. I even tried taking the drives from the working machine and putting them in this one. The problem stayed with the hardware.

After attempting BIOS upgrades, BIOS downgrades, system partition re-creates, etc, I finally broke down and did a systematic line-by-line comparison of BIOS settings to see if there was anything going on. There’s a BIOS setting labelled “Operating System”. On the working machine, this was set to “Windows 2000” (I never bothered changing the setting from when I acquired the machine). On the broken machine, it was set to “Other”. I have no idea what deep underlying changes this setting makes, but for Fedora Core 5 on a Compaq DL360, it needs to be set to “Windows 2000”. Crazy.

Bell Doing the Right Thing

As a followup to my recent posting about Bell violating the GPL, I received a followup call from Mark at Bell. He said that the issue had been sent up to their legal department, and the decision has been made to publish the GPL’d code on the Bell website. mark told me that it would be the same code as published on the Dish site, and that it should appear within a week. As soon as I spot it on the site, I’ll post a link.

I am a bit suspicious that they’re going to post the same code as posted on the Dish site, because I believe that since the original Dish posting, there has been additional work performed on GPL’d areas, such as the SATA drivers. Once released, I’ll do a diff to see if it’s indeed the same code base. The best way to tell if Dish and Bell are being above board is to extract an image from the 9200 and compare strings with the released source code to see if there are any obvious differences.

True Alternate Worlds

I came across a very interesting read today. Laukosargas Svarog has used the framework provided by Linden labs’ Second Life and built what appears to be a true ecosystem. She has weather, birth and death of plants, animal polination. It’s a testament both to the flexibility that Linden has built in to their system, as well as the creative imaginations of a single individual. I wonder what would happen if her virtual plants were allowed to ‘escape’ her garden and colonize the entire world…?

Fedora Core 5 on Compaq Proliant DL360

I recently spent a whole bunch of time banging my head against a wall trying to get Fedora Core 5 installed on some Compaq Proliant DL360 machines I have. Core 4 installs fine, but Core 5 has a problem with the hardware RAID on these machines. The installer fails to identify the driver (cpqarray) successfully, and if selected manually, it finds it but fails to find any hard drives attached.

After quite a bit of searching, I found a posting at the Fedora Forum describing the fix. The trick is to boot the installer with:

linux noprobe noapic noapci

Once the installer fails to find the hard drive, select ‘cpqarray’ and ‘sym53c8xx’ drivers. After that, the install will complete successfully. All credit to this fix goes to ‘Mesu’ from Fedora Forums.

Bell Violating the GPL?

I had heard a rumor that the Bell 9200 PVR used a Linux kernel. I haven’t been able to confirm that rumor. However, I have found confirmation that the Dish 942 is Linux based. I believe that the 942 and 9200 are basically the same hardware.

Dish has done the right thing and released portions of their source code because they’re using software that is licensed under the GPL. Their disclosure, although difficult to find, is here:

http://www.dishnetwork.com/content/products/receivers/dvr/SourceCode/

I’m going to give Bell a call and see if they will release the source code to me as mandated in the GPL. I’ll let you know how it goes!

Update 2006/05/25

I called Bell after posting this entry, and was passed on to Sharaz (LXY 6009835), a second level technician who seemed to understand the concept of the GPL. He was unable to locate any reference to it on the Bell site or in their Intranet. He assured me that he would look into it and call me back.

Two days later, I haven’t heard from Sharaz, so I called back. This time I got Mark (LJZ – he refused to provide his employee number. I’m not sure of Bell’s policy, but this is the first time that I have been refused this information from a Bell employee). Mark seemed to take an interest in my question. He sounded like a bit of a Linux nut himself. Hi appeared to be using an Instant Message client to talk to the development staff, but the response he got back was “What’s a GPL?”. Based on that answer, he decided that an email to the corporate side of things might be in order. He’s assured me that he will get back to me by Monday with a status report.

If I don’t hear back on Monday, I’m going to take the GNU project’s advice and advise the copyright holders of this apparent voilation.

I’ve found some interesting links discussing Dish and GPL violations in general:

Slashdot article about Dish releasing the code

Newsforge article about dealing with GPL violations

Harald Welte’s blog discussing GPL violations

ZDnet article about GPL violations

Harald Welte’s website dedicated to GPL violations

Be Careful What You Ask For

Our company has just gone through a merger with a US company (we’re Canadaian). So far, it’s looking like the majority of the combined company’s IT work is being done out of my office. The other day, I just received a document from the US office with some notes regarding an upgrade to their website. I’m getting rather used to someone handing me some scribbled notes on a piece of paper and expecting me to interpret those notes and produce a working piece of software. I wasn’t quite expecing these notes, however. The gentleman had literally taken the time to take his hand-scribbled note and scan it in, then convert it to a PDF, then email it to me. It’s completely unintelligible.

T520060509121050408

A Complex Satellite Distribution System

My problem was this. 5 Bell ExpressVU satellite receivers. One Free-To-Air (FTA) receiver, primarily used for receiving NASA TV. In order to feed all this, I have two satellite dishes. The first one covers 83° and 91°. The second one covers 119°. I need to get the Bell satellites to 5 receivers, and I need to get all three satellites to the FTA receiver. Here’s a picture of my distribution system:

T520060423202128718

t’s a little unclear to follow in the picture, so here’s a schematic version that is a little clearer:

satwiring

I have two SW44 switches to allow more than 4 ExpressVu receivers. The two SW44 switches are connected together using satellite-grade (up to 2GHz) splitters. The splitters I used have one DC path. Make sure that the DC path outputs are all connected to the same switch (it doesn’t matter which one). The splitters then connect to a pair of classic 2×4 switches (sometimes called 3×4 because there’s a third terrestrial signal input, which isn’t used). Outputs 1 and 2 basically pass on the 13v and 18v outputs from the LNBs on the ExpressVu dish. The third output from the switches, along with the line from the 119° dish connect to a DiSEqC switch. The DiSEqC switch then connects to the FTA receiver.

That’s it! 46 F connectors and a box of wire later! Make sure you use good quality connectors and cable. This many switches and connectors will degrade the signal strength, so you need to make sure you’ve got as little loss as possible. In my case, the signal strength dropped from around 90 to just over 80. Still very acceptable.

Bell 9200 Reaches New Lows

As I blogged earlier, there are continuing problems with the Bell ExpressVu 9200 PVR. This morning, I woke up to the only problem on the growing list that I had not personally experienced yet. Error 0521. It means that the hard drive is corrupted. I lost all of my recorded shows (which, because of the stuttering problem, were almost unwatchable) and all of my event timers (which, because of the lack of name-based recording, were at the mercy of the networks shuffling around show times). I called Bell, and spent several minutes re-explaining all of the issues with the 9200 PVR, to which they agreed. They are shipping me a replacement unit which should be here in 3 to 7 days. I seriously doubt that the replacement is going to fix anything at all. Infact, I believe that the replacement shipping is merely a stalling tactic to try to delay another week or so until they can finally get all of the bugs out of the fabled software update.  I’m not the only one with problems with the 9200. I think Bell has a major PR problem on their hands, and they had better do something to fix it quick.

As a quick recap, here’s the list of 9200 problems that are known:

  • Stuttering playback of recordings
  • Green pixelization which doesn’t go away until unplugged
  • Name based recording missing
  • Caller ID not working
  • Error 0521 causing all hard drive content to be lost
  • Interactive TV features not working

Kenwood TM-D700A Repair

A fellow ham approached me with a defective Kenwood TM-D700A transciever. The unit showed no signs of life. There was no display, and no response to pressing the power button. The only hint was that the cable between the radio and remote head had recently been repaired.

Luckily, I have the service manual for this radio. It was essential in this repair. After checking the obvious (making sure that power was getting to the radio), I decided to check that power was getting to the remote head. In order for the soft power switch to operate, there is a constant supply to the remote head. There should be 9.66v between pins 1 and 4 of the remote head cable. In this instance, there was no voltage at all. A simple power supply around D903, Q910, Q911 and Q913 creates the 9.66v signal from the raw 13.8v. Again, no voltage at all. The supply is protected by fuse F902. This is a 1.8A tiny surface mount fuse. Without the schematic, I probably wouldn’t have found it. It’s located just below the voice synthesizer connector (see rather bad camera phone picture). I was unable to locate such a tiny surface mount fuse, so I replaced it with a 2A picofuse. I carefully bent the picofuse leads to attach to the original solder pads. A small dab of glue gun glue held the fuse up off of the circuit board to avoid any shorts.

After re-assembling the radio, all functionality checked out. I suspect that the original damage to the head cable caused a short which blew this fuse.

T520060428170625276T52006042818562153T52006042819462673

Disassembling the 9200 PVR

Never one to shy away from voiding a warranty, I decided to take the plunge and pull apart my malfunctioning 9200 PVR to see if I could figure out if the hard drive had any sort of AAM (Advanced Acoustic Management) enabled.

Opening the case was easy enough. It consists of removing four screws holding the top cover on. The cover on my unit was a little tight, and needed some gentle persuasion. Once the cover is off, you’ll notice a small silver sticker on the hard drive bracket. This makes it impossible to remove the hard drive without breaking the sticker and voiding the warranty. Luckily, drive removal was unnecessary because I was able to remove the SATA cable and power cable from the drive, and plug in cables leading to my desktop PC.

Once hooked up to the PC, I was able to recognize the drive in the BIOS. It comes up as a ‘MAXTOR SABRE’ drive. At this point, I tried several utilities to see if I could detect and adjust any AAM setting on the drive. I tried several DOS-based utilities (the Hitachi utility is supposed to work with many manufacturers’ drives, but didn’t work in my case). The only utility I was able to make work was a utility called Doc’s AAM Utility. Unfortunately, this utility is in German, and my German is non-existent. It appears that the drive was recognized, but that AAM was not supported on the drive. As of yet, I’ve been unable to confirm or deny whether AAM is a feature of this drive.

It’s late, so I decided to put the PVR back together and call it a day.

T520060417221559934T520060417234235958T520060417234247988T520060417234300721T520060417234314376