Ninety-ninety Rule of Project Schedules: "The first ninety percent of the task takes ninety percent of the time, and the last ten percent takes the other ninety percent."

In the Trenches with LAROKE

Konsultant's Log, Cyberdate 05.13.1998 (Building Johnny Mnemonic Part II)

SITREP: At the end of our last log entry, the server PC "Johnny Mnemonic" had been fired up for the first time. He is in a raw state. The hard drive is not even formatted yet. We have yet to do that as well as optimize BIOS settings, install operating system software, network adapter, NOS and other devices. The important thing at this juncture is that the machine is up and running without odd noises, sparks or smoke.

TACAMO: Johnny was rebooted and the BIOS setup entered again. I used the "Reset" button on the front of Johnny's case to reboot . . . Good! Guess I got that one attached right too.

Task 1: Configure BIOS for system setup

Following the PC Guide's recommendations, I proceeded to configure the BIOS for "safe" low-risk settings to get the bugs ironed out if there were any. This is also the reason I have not yet installed the SCSI hard drive and controller, or network adapter.

First, the clock was set for proper date and time. The "IDE HDD AUTO DETECTION" menu item was used to detect and enter the settings for the Seagate drive into the BIOS for Primary Master. I left the Secondary Master on "AUTO" since there was not a setting for "CD-ROM" I changed Primary Slave and Secondary Slave settings to "NONE".

In the "BIOS FEATURES SECTION" I disabled the "Quick Power On Self Test" so that all POST tests would be performed. The "Boot Sequence" was changed so that drive A: would boot first, then C:. Video BIOS Shadow ROM was disabled.

"Power Management" was "Disabled" in the "POWER MANAGEMENT SETUP" section. In the "PNP AND PCI SETUP" section, "PNP OS Installed" was changed to "Yes" for the upcoming Windows 95 installation.

In the "CHIPSET FEATURES" section, the "Parallel Port Mode" was changed from "ECP+EPP" to "EPP" to save on DMA resources.

Murphy makes an appearance

The BIOS setup was saved and exited. OOPS! after rebooting, I got "CMOS Checksum error Defaults loaded" followed by "CMOS battery failed" error messages. All the above settings were wiped out.

UHHHMMM . . . I entered BIOS Setup again and reset some, but not all the changes I had made above, saved and rebooted . . . no error messages this time, but I had an uneasy feeling. When I had set the system clock, It had not been off by minutes, hours, or days but two whole years! Maybe the battery was going bad. This would be the time to replace it before everything was set up and fine-tuned. I shutdown Johnny and cut his power off at the surge protector. later when I boot him up again, I'll check the system clock for time loss.

I checked the depleted computers "Krash" and "Christine" on the off-chance they used the same CMOS battery but they don't. The one nice thing regarding CMOS batteries in newer systems is that they use the CR2032 3v microlithium coin cell batteries commonly used in watches and calculators in lieu of the hard-to-find speciality batteries of the good ole days. I had to pick up some memory for another system anyway, so I left the office for a memory and battery road trip.

2:49 P.M. 4/17/98 When I got back from the road trip I started Johnny up again . . . He was holding date and time OK, and I did not get the previous error messages. I had one other error message to deal with at this point. It was displaying at the end of the bootup process, but it was not stopping the system so I had ignored it until now. The irritating, but non-alarming message was "BIOS update data incorrect (CPUID=00000634). Update not loaded."

The motherboard manual mentioned that the first time in BIOS setup the "LOAD SETUP DEFAULTS" option should be selected before making system changes (after making a BIOS copy on diskette with Flash Memory Writer utility). Saving and exiting BIOS setup revealed that the message was still there . . . probably time to visit the motherboard manufacturer's Web site. I was also beginning to wonder if this message had anything to do with the enabled "Boot Block Programming" jumper I had found earlier when checking the motherboard jumpers

At the ASUS Web site I downloaded the latest BIOS update file for Johnny's motherboard and unzipped it to the boot diskette. Reading the instructions indicated that if power is lost during a BIOS update, the motherboard will become unbootable and must be sent back to the manufacturer. To avoid this possible catastrophe, I determined it was time to setup the UPS that was purchased to use with Johnny. After that was done (and tested), Johnny was restarted and the Flash Memory Writer program was started at the A:> DOS prompt again. The BIOS update downloaded successfully into Johnny's BIOS and the machine was shutdown and restarted. The BIOS setup was re-entered and the "LOAD SETUP DEFAULTS" option selected again as instructed by the motherboard manual and the Flash Memory Writer utility's documentation file. The BIOS settings were saved and Johnny was rebooted again . . . Darn! The error message did not go away.

I entered BIOS setup again where I discovered a "BIOS Update" setting in the "BIOS FEATURES SETUP" section. This setting was enabled by default. The manual stated:

"This functions as an update loader integrated into the BIOS to supply the processor with the required data. The BIOS will load the update on all processors during the system bootup in the default position of enabled."

Well, not sure of what this means, but I decided to see what disabling this setting would do . . . When Johnny rebooted, the annoying message had disappeared. I'm going have to research this setting, but for now I'm going to leave it disabled.

12:52 P.M. 4/18/98 Today when I started Johnny up for the first time everything was OK, but at one point when I rebooted after turning the power off at the UPS, I got the nasty "CMOS Checksum error Defaults loaded" followed by "CMOS battery failed" error messages again. OK. Enuf of this. I replaced the CMOS battery with the new one I had purchased yesterday. I tested the new battery over the course of the next half hour by turning Johnny on and off for various periods of time after adjusting the BIOS settings for the third time. I turned Johnny off both by using his switch (normal shutdown), and the switch on the UPS (simulating a power outage). I did not get the error messages again and am cautiously optimistic.

Time to setup the Seagate hard drive. Following the excellent instructions on the PC Guide Web site, I used the Windows 95 OSR2 versions of FDISK.EXE and FORMAT.COM to setup the Seagate drive as one large FAT32 volume and labeled it "Charlie" (see LAROKE drive naming conventions).

Task 2: Install SCSI Devices

Next the CD-ROM drive was setup with its DOS drivers so that software on CD-ROM could be installed, namely, the software for the ADAPTEC AHA-2940 SCSI controller that would be installed next, along with the Western Digital 9.1GB SCSI hard drive. Again, the instructions on the PC Guide site makes this painless and can save you a multitude of "gotchas".

2:26 P.M. 4/18/98 Johnny was shutdown for the installation of the SCSI controller. The controller needs a Bus Master PCI slot. As I could not find specs regarding this issue in the motherboard manual, it was cause for concern. I may have performed another faux pas here by not confirming this spec prior to purchase. Nothing for it, but to find out the hard way. I installed the controller in PCI slot 1, my best guess for a BUS Mastering slot.

After checking all the Adaptec documentation, I determined the defaults were OK and no changes had to be performed prior to installing the Western Digital drive. The 9.1GB SCSI drive documentation was checked and the default jumper settings were aceptable, so the drive was installed in the bottom bay of the removable "mounting box" that the floppy drive had been installed in earlier. power and internal SCSI ribbon cables were attached. The large power connector was the type needed, and since the SCSI drive was located far from the power supply, the only connector that would reach was the auxillary fan pass-thru connector installed earlier. The 68-pin (ultra-wide) internal SCSI ribbon cable that came with the Western Digital drive was about six inches shorter than the one that came with the controller, so it was used since the controller and the drive are relatively close to each other.

Johnny was booted again, and controller and drive were found without problems . . . Things were going too well, and I began to look over my shoulder on occasion. FDISK.EXE and FORMAT.COM were used to get the biggest possible drive which, under OSR2 FAT32 turns out to be 8.65GB since Windows 95B cannot handle a 9.1GB drive. Since the Windows 98 version of FAT32 is supposed to get by this limitation, I decided not to clutter things up by trying to install overlay software to get the whole 9.1 gigs at this time . . . I plan to update the drive when Windows 98 can be purchased and Johnny can be upgraded. After all, I'm trying to get away from the situation I have on the current file server "Old Blue" which has a combination of 3 Stacker compressed partitions and two DOS partitions, all FAT16. The capacity of Old Blue is six gigs spread over five drive letters and it's becoming a file management nightmare.

"Surprise! Surprise! Surprise!" Gomer Pyle would say. I found out that there was no cause for concern since the 9.1GB drive turned out to be 8.46GB actual when the whole drive was formatted, the difference being attributed to "marketspeak" in much the same way that monitors with 15.5" diagonal viewable screen area are sold as 17" displays. A Gigabyte is 1,073,741,824 bytes whereas Western Digital defines it as an even 1,000,000,000 bytes.

Task 3: Install the Network Adapter

Johnny was shut down to install the Invisible LAN NIC cannibalized from the PC known as "Christine". It was installed in ISA slot 1. Johnny was connected by thin-coax to the company's 10Base-2 Ethernet segment that snakes through the building and now acts as a low-bandwidth "backbone" connecting individual PC's and 10BaseT hubs to the network. For this reason I've named this thin-coax segment "The Stillwell-Burma Road" in honor of my dad who served in the CBI theatre in WWII and also to reflect its low-bandwidth convoluted path thru the office (overhead in the ceiling and walls in some places, underground and in conduit in others). Our first 10BaseT hub has been named "Deep Space 9" and the next one will be "Babylon 5".

After Johnny was restarted, The DOS utility that came with the NIC on diskette was started. This Utility can view and change settings for the adapter which are stored in an EEPROM. This is not as good as PnP but much better than jumpers and dip switches. The NIC's existing settings were for IRQ3 and 300h I/O address. I decided to change to IRQ5 to avoid conflict with COM ports. The new setting was saved to EEPROM and the utility exited. Johnny was reset and the BIOS setup entered. In the "PNP and PCI SETUP" section, I reserved IRQ5 for ISA slot use to prevent PnP from "grabbing" for another device during bootup.

Task 4: Install Windows 95

4:30 P.M. 4/18/98 Time to try for a successful Windows 95 installation. Windows 95 OSR2 installed cleanly, without any conflicts, from CD-ROM. I was on a roll here. A custom installation was selected and a minimum of components installed in keeping with Johnny's role as a server. All devices were found and generic drivers installed for them. Reviewing the Windows 95 "Device Manager" showed there weren't any conflicts. I reviewed the LAN adapter's resources specifically, which revealed that Windows 95 had chosen IRQ3 instead of IRQ5, but I was permitted to edit the setting to IRQ5 to put things right. No other changes were made. The setup program had installed Microsoft Networking and the PCs "4-bits" and "HAL 9000" could see and access Johnny in the Windows 95 "Network Neighborhood".

CM: 9:51 A.M. 4/19/98 First order of business today: fine-tune the Windows 95 setup. In the Device Manager, Under the CD-ROM's Properties, the "Autorun" feature was disabled. A "default" printer was added to the Windows 95 "Printers Folder". There would not be any printers connected to Johnny physically and I would not be installing any network printers until the Invisible LAN NOS was setup, but one printer every Windows 95 system should have installed is the "Generic/Text" printer. This is a printer driver supplied with Windows 95 that will print a "generic word processor" text document. I always set it up to print to a "FILE" instead of a printer port. This arrangement allows me to print documents to a file that word processors can edit (with their default "PRN" extension) and many other applications can view, including Web browsers (if you name them with a "TXT" extension).

The Windows 95 default performance settings were then tweaked with assistance from the PC Guide and PC Mechanic Web sites. In the "File System Properties" Dialog the "Typical role of this machine" setting was changed from "Desktop computer" to "Network server". In the "Virtual Memory" Dialog, the "Let me specify my own virtual memory settings" option was selected, and "100MB" was entered for the "Minimum" and "Maximum" swap file size. This would force Windows to create a permanent 100MB swap file instead of constantly resizing the swap file which is a performance drain. The general rule-of-thumb is a 50MB file but this is a server and I have plenty of room on this drive. Next, the Windows 95 Disk Cache was limited in size to 8MB and prevented from resizing by editing the SYSTEM.INI file in accord with the methods outlined on the PC Guide and PC Mechanics sites.

The next task was to install updated drivers for the installed peripherals. The CD-ROM drive did not have any specific Windows 95 drivers so that left the Adaptec SCSI host adapter as the only item to install drivers for. A 3-1/2" diskette and CD disk were provided. The setup utility ran from the floppy drive and installed components from the CD. In addition to drivers, I installed the included monitoring and utility software but none of the "Multimedia" components.

The only elements left to complete the basic setup were to finish the hardware loose ends and to optimize the BIOS settings. Johnny was shutdown and his case removed. Final visual inspection and connector checks were made and serial numbers recorded for devices I had forgotten to document earlier. Johnny was buttoned up and restarted. The BIOS setup was entered. In the "BIOS FEATURES SETUP" section the "Boot Sequence" setting was changed to "C,A" and the "Boot Up NumLock Status" setting turned to "Off". These were the only changes made at this time. The performance settings that would affect a user if this were a desktop system were not enabled because of Johnny's role as a server. "Power Management" settings will be tweaked later when the UPS monitor connection and software is installed (and I learn more about these settings).

MISREP: Johnny was basically set up at this point and communicating with some of the other machines on the network via the Microsoft Networking protocol. The operating system was installed and he was now ready to start assuming the duties of Old Blue, the current File Server/Web Server. Configuring Johnny as a server, moving data from Old Blue, and reorganizing the file structures will be the topic of the next "In the Trenches" installment, along with some final hardware additions. Then, Old Blue will get his makeover as a communications server.

LAROKE Microcomputer Consultants
155 East Boca Raton Road
Boca Raton, Florida 33432
(561)368-0659 (Tel & Fax)

Issued Wednesday May 13, 1998

copyright © 1996-1998 LAROKE Microcomputer Consultants all rights reserved