Entry #1a: The build begins
Frankendate: 07.19.2006
Frankie III has arrived at my house. Thus far nothing has happened
with him other than moving him to 'the lab'.
The 5075 tower has been opened up as has the 7104 expansion unit. As
expected the 7104+5075 combination is exactly the same size as a
270+7104 or 810+7104. So setting them side by side and setting the
new wider top on, it fits perfectly. So cover wise we may end up OK.
Certainly sides and tops will fit. Front is an issue.
Now the 5075 opens on the right instead of the left as a 270/i810
does. Meaning access to the Right side is critical. The 7104 is
intended to be mounted on the right side and access to IT is also
from the right side.
The metal housing of the 7104 is chiefly once large piece of sheet
metal that starts out as the bottom, makes a 90 degree bend and
forms the entire back side (meaning middle of the thing once
installed) and then another 90 to form the top - a large block 'C'
i.e. '[' Then a piece is riveted on the front and back. The open
side is the black sliding cover.
So one potential option is to mount the 7104 on the left, but mount
it backwards. This would push air back to front, which is a problem
because... Um, well it's not. Yah the power plug will be in the
front but this is frankeni.com after all!. The other minor
issue is the tab which is supposed to stick out the front to hold
the front cover. It if course, sticks out the back that way. It
might be able to be moved if it turns out to be a problem - drill
out some rivets and bold it to the front.
It's possible to mount it backwards, i.e. the open side 'In', still
on the left side but there's virtually nothing in there that I need
access to once it's working.
Next is the issue of getting the thing to power up. The 7104
requires a 20 pin cable that connects daisy chains into the debug
port on the 270. The ribbon cable that goes from the main board in
the 5075 to the 'control panel' of the 5075 is this same exact
format. I'm not yet comfortable with that working but it might! I
need to do some electrical analysis of the signals on that cable. A
second signal comes to the DASD cages to light those buggers of. I
believe all that is is a parallel to the one in the main cage. Again
I'll need to check that but this one should be easier. There is a
little pigtail in a 'narrow' 270 that this connects to but I didn't
see that in the 5075.
Other items I'm gonna mess with. I intend to attempt to put a
non-RAID card from a 270 (the base 4 drive only controller) into a
slot next to a tape card, connect that to the internal DVD, make the
pair switchable, and then hope this makes the DVD switchable!
Clearly something the LVT would scream about but hey, this is
Frankenseries.
Depending on the way disk works out I have the FrankenTower Mini
(two drive cages) that I could use to house 10 more drives as well.
As I'll be heading for vacation on the 20th, build time until August
will be virtually zip.
- DrF
|
Entry 1b:
Frankendate: 08.02.2006
I have made small progress. The V5R2 load currently on him has been
shrunk to 6 8GB disks. It will be moved to a partition when I get
that far.
Another thing is the use of a FC #2767 RAIDless IOA (the base IOA in
small 270s) does work for the DVD. I stuck it in an open slot,
re-cabled the DVD to it instead of the RAID card and it came right
up as OPT02. So I think that idea of making it switchable with the
FC #2749 card will get us a CD/DVD for LPARing the thing.
The cable from the Planar on the FC #5075 is NOT the same as the
debug cable so it cannot be used to power up the FC #7104. I'll be
checking out alternatives for that.
Anyone know if you can put FC #4326 15K 25GB DASD in the base
6-drive can from a FC 7116 (which goes on an i800/i810?) I intend to
swap that into the base of the 270 because the base 270 cage is
listed all over the place as NOT supporting those drives.
- DrF
|
Entry 1c:
Frankendate: 08.02.2006
I have run into a hitch (for the moment) in that I cannot get the
expansion unit to power up on it's own. When connected to the 270 in
place of it's sidecar it DOES power up so I knew it's good. There
are only 6 pins in the cable that are in use.
1-Ground Hard to get that one wrong.
2-Fan speed return to the 270 CEC for the cabinet fan, Verified this
one by pulling the sidecar fan and the 270 goes down with a power
code indicating 'air movement device failure'
3-Fan speed return to the 270 CEC for the power supply fan. Haven't
verified this because the fan is inside the PS but it behaves
exactly as the pin above.
4-+5 Volts from the expansion unit. It's present as soon as the
power supply is connected to the wall. (Same as expansion on the
270) When connected to the 270 this falls to 4.18volts. It does not
change when the 270 (and sidecar) power up.
5-No voltage when the sidecar is by itself. Gets boosted to +5V when
the 270 is connected. Does not change when the 270 and sidecar power
up.
6-No voltage when the sidecar is by itself. Gets boosted to +5V when
the 270 is connected. Drops to 0V when the 270 spins fans and lights
up drive cages. Even when I try to swing this from +5V to Gnd
nothing happens.
So a temporary impasse for sure but frustrating nonetheless.....
- DrF
|
Entry 1d:
Frankendate: 08.03.2006
Latest update is that I DID get the second sidecar to fire. Turns
out the one pin that seems to be only a 5V or 0V is actually digital
data. So I connected the Gnd, +5V, That floating pin and one other
one at the same time as the sidecar on the 270. This powers up BOTH
sidecars. Woo Woo!
It does yield an SRC 1011 1522 on the front panel but everything
runs. The SRC says P02 power supply fail (expansion unit) but I will
explore some more and see if once the 'extra' expansion is up
whether I can disconnect the extra pins without powering the thing
off and thus avoid the SRC. The SRC undoubtedly comes from TWO power
supplies talking back to CEC and confusing it.
- DrF
|
Entry 2a:
Frankendate: 08.21.2006
Frankie III made another step tonight. He has fired up, sorta. I
eventually did give up on lighting both 7104s from the CEC. I
couldn't get that to be reliable and hence plan "B" was hatched. I
pulled the IBM power supply and grafted in (it fit quite nicely
actually) the PS from a big Compaq rack mount DASD box that I just
scrapped out. That PS drove 8 'full high' drives so driving 12 half
high units shouldn't be a problem. Only issue is it needs to be
turned on with a switch. Oh well.
I also succeeded in grafting the 7104 onto the 5075. That turned out
too sweet for words! The thing fits perfectly, albeit it needs to be
removed in order to play with cards in the 5075 since it's on the
right side covering those cards. Yah well, that way the top front
AND back covers fit perfectly. Sweeeet!
Now I made a tactical error in the build because the 6 drives in the
CEC now are supposed to move to the 2nd 7104 and be a V5R2
partition. They will be driven by a 4778 RAID card (and were while
in actual production.) Now this is no problem except that since the
7104 covers the 5075's slots I moved things around and the 4778 is
already 'upstairs' and I put the 2780 downstairs. So at IPL time the
card (I believe) is re-striping those 8GB drives '2780 style'. I
think this because a) They are all 6 going zonkers and b) I have an
SRC B164 4605 which indicates 'bad attitude on the part of the load
source drive.' So I'm letting that cook until the drives calm down
or 2 hours pass because by then they are either a) Done converting
or b) Oh shit.
Pictures tomorrow.
- DrF
|
Entry 2b:
Frankendate: 08.21.2006
Well after converting the drives to the other (wrong) RAID stripe I
bounced him a nd up he came. Life is good except that now I know I
need to end RAID to put those drives back on to the 4778 card. Such
is life. By the way anyone know how to tell which type of RAID strip
the drives have? Or do you just infer from the previously attached
RAID card?
Attached are some pics. Front and back of the 5075+7104 and the
entire stack. Right now it's in an incredibly bad spot so I gotta
fix that but it can be easily accessed there, likely the only plus.
Note the single drives in each can, that's just so find out if
things are working or not and they are. The drives in the 5075's
7104 report 'unresolved locations' for some reason. Hmmm. :-)
Today I hope to try to get V5R4 onto the primary partition, I'm
going all LAN console top to bottom (Hence all the ethernet cards
you see in the 5075, useing a few FrankenRules there!) Note the
power switch hanging at the back of the 7104. That goes to the
Compaq power supply on the bottom. See how nice that fit?
Incredible. The black strip on the bottom is Duct tape covering the
1/4" slot it left so that air flow is correct. I think I'll be OK
with cooling because though small the fan in that PS is one
agressive mother. (Shades of the Indianapolis Cert Lab!)
My goal is to have this thing working and pics up on
Frankenseries.com B4 COMMON. So Kev plan on a little last minute web
work if you can.
Gee I'd love to take FrankieIII to Miami but it would cost a
freeking Shitload to ship him and it's just too dang far to drive
although at 1,500 miles in my car it's less than $100 in gas. Now if
Jim were to fly over here and we shift drove. Naaah it's 30 hours of
driving. Each way. And we're in Anaheim in the spring. (2200 miles
and 36 hours). They ain't making this easy!!!
- DrF
|
Entry 3:
Frankendate: 08.21.2006
Hmm. I seem to have hit an interesting roadblock. I'm working the
installation of V5R4 onto the primary partition. This has a FC #2780
RAID card, a FC #7136 DASD can from a FC #7116 and 6 FC #4326 disk
units. It also has a dozen FC #4318 17GB Units in the FC #7104.
The Tape drive (50/100GB QiC) and DVD-RAM drive are driven by a FC
#2742 and the base FC #9167 from a 270 I upgraded some time back.
These will be switchable once we get that far (and the FC #2749
that's with them.)
First issue is the IPL from I_BASE (RSC) takes a freeking long time.
Much of the time is spent with C1F7 280A and C167 280A alternating a
few times a second. It spends like half and hour on that one. I
can't find a reference ANYWHERE to what those codes mean. In any
case eventually it does move on. C600 4508 also hangs on for a long
time unless you do F21 right when the code pops up.
Now here is where I have a problem. When I finally get the screen to
install the licensed internal code it chokes. It complains there is
a problem with the Load Source disk unit. It shows all the important
stuff but lists the drive as a FC #6600 instead of FC #4326. If you
proceed into the limited DST environment provided by the CD you can
see them all in there and all 6 of them are FC #6600-050s. Looking
at the detail they report in as part number 97P3030 which is correct
based on a few other systems I checked with FC #4326 disks.
In scouring system builders and other doco there is a note that I
found that FC #4326 disks are supported by 270s only in FC #5095s.
Now I expected that meant it was a cage sort of thing. FC #4326's
are supported by i800 and i810s in both the CEC and the FC #7116 and
FC #7136 expansion for that. Those expansion CASD Cans report in as
CCIN #28CD. WHen I look on FrankieIII I see that can reporting CCIN
#28CD. The FC #2780 driving those units seems plenty happy. The
remaining cans report in as CCIN #283F.
So now I'm wondering if 270's can see FC #4326s in FC #5095s and
i810s can see FC #4326s in FC #7116s and the FC #2780 is the best
card for those drives no matter where they live and if any OS can
see them V5R4 should absolutsly see them then what could be missing
that makes them look like 6600s to SLIC?
So I flipped two SCSI Cables effectively putting the FC #4326s into
the expansion and making #4318s the load source I loaded SLIC to
that and IPLed to DST. The disk units still show p as FC #6600
though they DO show up as 35GB Units in Work with Disk Units. If I
attempt to nitialize and format them in disk unit problem recovery
procedures then I get an instant error bout there not being support
for those units in the system. Boo.
I have another theory that the IOP is the problem with the big
drives. I swapped in a FC #2843 for the FC #2842 to see if the
system gets all cranky with that. It is happy and doesn't seem to
care a whit about the FC #2843. My guess was the FC #2842 was
created with less memory in theory simply to be cheaper. You wanna
use a 'more expensive' IOP? That's OK. Now however FC #2842s are
more expensive on the used market since they are comparitively rare.
So the next theory is that I may not absolutely have to put the
cards in the designated slots. OK that's not a theory that's a fact.
My theory is that support for these FC #4326s things is actually not
in the DASD IOA/DASD can combination, rather it's in the IOP. Now if
that is the case then If I were to put the basic SCSI card in C01
where the big RAID card is now and drive tape and DVD with that,
leave the two line wan/modem in C07 and put the FC #2749 in C06 then
I have achieved switchability as I want. I then put in a FC #2843
which in theory would support the 2780 with the big drives.
Well the theory may be OK but in practice it's been a bitch. I
absofreekinglutely CANNOT get the 2780 to stick to the 2843! No
matter what set of slots I use the 2780 gets owned by the MFIOP! I'm
not kidding, you put that 2843 in slot C05 with the 2780 in C04 and
the two 2838's (Console and LAN) in C03 and C02) and the damn 2780
shows up under the MFIOP (EMB)!! Move the 2843 up to C06 and the
same thing. Move the 2780 up to C05 to be right under the 2843 and
same freekin thing! It's nuts! I never thought the silly thing could
puls slots in the middle of the bus onto another IOP I always
thought it was sequential with all IOAs from one IOP to the next on
the same IOP. So much for THAT theory.
There is some hope that when building the LPARs I might be able to
make it work yet but we'll see after PTFs are applied and the system
is switched to shared processor mode.
- DrF
|
Entry 4:
Frankendate: 08.22.2006
FC #4326 update.
IBM was sure there were reasons that the 4326's won't work in the
CEC. They were quite certain they got tested in there but drew too
much power and made too much heat. Also quite adamant that running a
big stack of them in the CEC wouldn't be a long term good idea. They
couldn't remember a specific block in there for those drives. He
thought maybe a FC #2844 IOP would be a better bet to support those
drives since the FC #2843 was around earlier (i.e. 820) before those
drives appeared.
I am currently attempting to trade the FC #4326s for at least as
many FC #4319s since performance isn't huge anyway.
V5R2 partition is up and running as it was the old 8.58GB Drives. I
re-raided them and they are happy. V5R3 partition is going as well
but so far just SLIC and basic i5/OS. My 8MM tape drive is all mad
about being dirty (and I have no AME cleaning tape!)
I also wonder if I can hang some more external DASD Cans on there
a-la Frankie I. Hmmm more things to try, more power to burn.
By the way my Wife (MrsFranken?) wants to evict the FrankenLab from
the basement. She wants me to add on to the back of the garage and
move all my stuff and my office out there. Now I like this idea a
lot I just need to get some $$s together and then call my friend
Dave the Builder. :o)
- DrF
|
Entry 5:
Frankendate: 08.25.2006
Tried something else today - putting in a FC #2741 to see if I could
make that switchable for a Switchable iASP. Get a nice neat SRC for
that one. 2741B934. Looked it up: 'Incompatible hardware detected.'
Got me on that one. I knew the book didn't list that card as
supported but what the heck. Now we know.
I am going to swap the 2741 from Lily into Frankie and swap it for a
2748 which IS a supported card. This card will then drive the
FrankenTower Mini that I built some time back which holds 10 drives.
These will become a Switchable iASP. They will switch with the tape
and DVD though so when connected to the V5R2 partition it's likely
they won't be recognized.
Got Domino running in one partition to replace Lily so that Lily can
begin her final journey. Need to pull out all the stuff that's mine
yet and that will wait a bit yet.
- DrF
|
Entry 6:
Frankendate: 08.28.2006
Well I powered down to put the FC #2748 card in the CEC instead of
the FC #2767 base SCSI card. I also stuck a half dozen 8GB drives
into FrankenTower Mini (Two DASD cans from a 720) It happily was
SUCKED into the MFIOP! This resulted in a very nice SRC 27809203
which when you ferret it out says 'What the? I can't IPL with the
system this way!' So I tried moving the card up a few slots so that
it is immediately below the FC #2843 card. This made it much happier
and the drives are now switchange along with the tape and CD. I'm
not sure how usefull this will be yet but we'll figure that out.
I also got the 10K RPM 35GB Drives (4319s) andstuck those in. They
work LOTS better than the 15K units. The V5R4 partition now has
368.6GB usable DASD.
So I get the system back up and tell it to IPL the V5R3 partition
and it tells me there is NO i5/OS license available. That's when I
realize that when I initially installed it I accidentally fat
fingered the year as 2004. Apparently correcting that to 2006 used
up about 730 of my 70 grace days. Dangit!
To anyway I'm now reloading i5/OS on that partition. Grrrr!
- DrF
|
Entry 7:
Frankendate: 09.07.2006
Well another fine success. After doing most of the cluster
configuration (I can't do all of it because I don't truly have
switchable hardware) I created a switchable iASP on the V5R3
partition. I created a library in it and a data area with text in
it. I then varied it off and switched it to the V5R4 (Primary)
partition. I varied it on and shazaam! There is the library and data
area! Sweeeet!
All does not go happily though,you can't switch this mess with the
iNAV gui. Because there is disk on the IOP, it gets pissy and says
'Sorry not a valid operation' Whatever. When you use SST to switch
it you get a stack of errors, one for every disk unit and one for
the 2748 card. Telling it you don't care it switches OK after that.
Cool! When you vary the thing on in the new partition you also get
about 100 messages but it switches!
More testing will be required of course and I also may dump the 2771
card and add another IOP so that the 7208 tape drives can be
switched separately from the disks, internal tape and DVD-RAM.
- DrF
|
Entry 8:
Frankendate: 09.13.2006
Found out something about this switchable iASP. Turns out that as
soon as I varied it on on the V5R4 partition it converted the thing
to V5R4. Switching it back yields a nice error about the iASP being
at a level that is newer than the V5R3 which is on the partition. So
much for that theory!
So I moved the FC #2748 to be sucked into the MFIOP so that the iASP
is now stuck to that partition. I may use it to hose the Linux
partition's disks.
Another FrankenDonor hit the lab today and tiny i820 with a fully
loaded (and monsterously heavy) 5035 attached. It has only two 17GB
10K Drives so I'm going to keep those as spares since there are a
fair number of those in FrankieIII. There are 8 of the old (and
monsterous) 17GB 7200 RPM drives in the thing and those may go into
the FrankenTower mini or they may go into Frankie II to support his
'PTF habit'.
These are about the last changes before COMMON as I'm severely time
limited between now and then and don't want him all busted while I'm
gone.
Interesting side note, our heat pump siezed in mid August so we have
no official furnace in the house. FrankieIII is keeping the basement
at around 70 even though it's been in the 50's and 60's around here
for several days.
- DrF
|
Entry 9:
Frankendate: 10.2.2006
Finally some time post COMMON. Not a lot, just some. I had created a Linux partition
before COMMON but had not had the time to crank it up. I worked on SLES9 for a while but
couldn't get it to pick packages. A Linux on i expert tells me I need a PTF for V5R4.
Still waiting for that. So I switched to RHEL and it came up perfectly. Currently I'm
using NAT through the primary partition and the virtual LAN with no issues. We were able
to utilize this for some testing at the Linux on i workshop over this past weekend.
Speaking of the workshop, thanks to Justin Haase for a donation to Frankenseries.com. A
very clean wide 170 with a number of usable parts. He was in RST on Tuesday, swapped the
thing to the 'Large Package' van who drove it to Chicago where I obtained it yesterday.
Unbelievably quick and cheap to get the thing from Iowa. It even came with a Twinax console!
Working on ideas to remove some other lessor utilized cards so that I can maybe get some
direct Linux I/O in there. Just for fun. I have my eye on a 2742 card that's not actually
doing anything but occupying a slot. Hmmmm.
- DrF
|
Entry 0Ah:
Frankendate: 10.7.2006
Frankie III is just too dang heavy! It's in a bad location and due to the 270 not having
wheels, well it's flat stuck there. So today a donor FC #5035 was stripped to the base. My
son and I removed most of 100 screws from the top 2/3rds of the thing and pulled that off.
This left the bottom section which normally holds power supplies, batteries, a UPS and the
lowest two drive cages and their fans.
Having this piece free we flipped it upside down and swapped out the straight casters with
a set of swivel units. Now it can be pushed in any direction and rotated as needed. I
figure the thing is strong enough to hold up the old red Wheelhorse plus the generator the
lawn mower the snow blower and a years worth of gas for the lot of them. As a consequence
I don't figure Frankie III will strain it at all.
Now thinking about the fact that it holds drive cans quite nicely and has a spot for fans
for those cages, hmmmm. Can you see more drives coming to Frankie III? :-)
-- DrF
|
Entry 0Bh:
Frankendate: 10.8.2006
Ah the Frankenbase. This is the name I gave to the new unit that will soon hold up Frankie
III. As mentioned in the last log entry it now has four swivel wheels to go in any
direction. I have also expanded the width of the lowest bays in order to mount drive cages
in all four slots. The next challenge is to get power to everything down there because
besides the drives, power is required to the fans needed to cool stuff down there.
Looking at the power issues it seems that this is virtually identical to the original
Frankentower! http://www.frankeni.com/gallery-frankentower.html We've got the same
power distribution available with cables that will power three cages and both fans with no
customization whatever. An easy add gets power to the fourth cage. Extending one cable
will also get power to a tray that holds three CD/DVD/ or full high tape drives such as
the QIC or 8MM variety that are so common in the Frankenlab. (Note, anyone with an extra
System i Ultrium drive of any size variety or flavor PLEASE Let us know!) I have plenty
of power supplies for this unit but they have one significant problem: They require 220v.
Of course the Frankenlab HAS 220v but I have no 220v UPSs available. It would suck to have
most of the system stay up but lose a bunch of drives every time the power hick-ups.
So looking at ways to use two 110v UPSs to get 220v. Shouldn't be a problem with the power
on as the UPS will lock frequency to the grid. As long as they are locked to the grid (and
connected to the opposite sides of the feed) there should be no problem getting a nice
solid 220v. Problem is when there is power there is, well power. I'll have to hook the
dual trace scope up to the output of the two UPSs and then kill power to both of them. My
guess is they will drift slowly apart eventually when they are out of phase by enough the
200v will fall below the minimum and down she goes.
-DrF
|
Entry 0Ch:
Frankendate: 11.19.2006
It's 4th quarter but the Frankenbase is nearly ready for deployment. It is now completely
wired. The PDU built into the UPS in there has been re-wired simply as a PDU. The fan that
was in the UPS is now wired to the power supply on the right side as have the other fans
in there. (A total of three, one large on the right, two small on the left. The SCSI
Cables run out the back and should be JUST long enough to reach the RAID cards which will
be above it in the 270 and 5075.
The Power connectors from the right side power supply have been re-worked to support four
disk cans instead of the stock three. My theory is the power required by the fourth can is
offset by the fact that we're not powering the big planar board in there it's three IOPs,
up to 15 IOAs and IPCS Card. I can't imagine all that doesn't add up to more than one
drive can. In addition the power to the removable device bay has been greatly extended so
that can be placed on top of the whole tower.
Hopefully I'll get some time over the Thanksgiving weekend to install this latest addition
to Frankie III.
Of course there is still that UPS experiment to perform....
- DrF
HO-HO-HO!
|
Entry 0Dh:
Frankendate: 11.26.2006
Well Frankie III is back on line though not completely happy at the moment. He is now
sitting on top of the new FrankenBase which, besides casters, is also providing power
distribution, room for 20 drives, and power for the external DVD-Tape (which is actually
the right side removable media tower from a doner 720.) Running on 230V he's drawing
right about 5 amps or about 1100VA. Let's not tell MrsFranken about that OK?
In order to make a second switchable tape controller I removed the production Ethernet
from the V5R2 partition. So that thing has only a LAN console port and a RAID card. How do
you get to it then? Well if Proxy Arp works for Hosted Linux partitions why not Os/400?
And so it is that the V5R4 primary partition provices access tot he V5R2 partition through
proxy arp and the internal Virtual Ethernet. Nice!
So the new stats for him are:
- 1070 cpw
- 4GB RAM
- 56 drives turning with several hundred GB of storage.
- Drives range from 8GB to 35GB.
- External 3490 switchable.
- Dual 7208 tape drives. Switchable.
- Internal DVD-RAM and 100GB QIC is switchable.
- External DVD-RAM and 6390 8MM tape, 'switchabe' (by moving cable!) This part is giving me
fits though. The dang thing makes the console port go away on the V5R2 partition! Hmmm.
- DrF
|
Entry 0Eh:
Frankendate: 12.28.2006
I learned some things about memory that annoy me. The RAM in Frankie III is DDR and is
variously stacked or unstacked 256MB Technology. So is the RAM in small i810s. A donation
of leftover parts from an i810 to i5 upgrade put a pair of 512MB and a pair of 1GB DIMMs
in my hands. The chips are identical near as I can tell so I loaded them into Frankie III
and fired him up. He was happy too and the DIMMS reported in with the correct features
codes even! Alas the size reported as zero. Thanks for that. So I pulled them back out and
am trading the 3GB i810 memory for 2GB of 270 memory with my favorite used guy. Oh well.
I'm having some issues with the DVD-RAM up top and it's tape drive buddy. They seem not
to report the correct SCSI addresses or something and freak out the partition. I've pulled
that off for now as I need the DVD for a customer install coming up anyway.
On another note the Big Red Dodge has been officially re-christened 'Old Red' as he's no
longer a daily driver even for the franken kids. Nearing 200,000 miles He'll ease into
retirement eventually but for now he's the utility truck needed to haul old systems and
firewood and such.
Oh and Frankie I is still missing. I'll pass along word through the frankennews service if
he's found.
- DrF
|
Entry 0Fh:
Frankendate: 04.07.2007
Been a while since the last update so several things to share. First off FrankieIII has
obtained some new tape drives. A donor 3590 leads the list. It's a big sucker too and the
autoloader hangs off the front of the rack! The next is a dual 8MM drive. These were once
the cat's meow, ok that was before my beard turned gray, but hey, it works!
Besides the new tape drives I have figured out how to address removable media. From donor
machines I have the removable media tray from both sides of a 720. The left side one holds
two removable media while the right side holds three. Now it turns out there is some
strange rule that the lowest slot in the right side needs to be a 2 byte (i.e. wide scsi)
device. Not sure what that's about but break that rule at your own peril. Now the cage
from the left side has an 'in' and an 'out' SCSI port while the right side has only an
'in' port with built in termination.
I hooked the stack to an open port on the RAID card. Unfortunately the corresponding slots
in those trays have identical SCSI Addresses so hooking them together means duplicate
addresses. In the IBM supported (read 'normal') configuration of these devices will have
them each on separate RAID controllers so the duplicate addresses are not an issue.
To overcome this I disconnected the ribbon cable in each tray from the devices (i.e.
Tape/CD/DVD) and using jumper pins, set the addresses myself. I IPLed the partition and
found my drives were quite upset and many not even working. Hmmm. Then I remembered a
friendly IBMer mentioning that for some reason the SCSI address is inverted. So No jumpers
(binary 000) would normally set the address at 0 but IBM inverts it to 7 (Binary 111.)
Armed with this information I re-addressed the drives and now have a total of 8 drives on
that partition plus two DVD-RAM Drives. Many options exist for tape conversions!!
Oh, also some of the drives were upgraded from 8 to 17GB 10K too. More space!!
Finally the new Tripp-Lite UPS is in place and working marvelously and has already
sustained two short power outages.
- DrF
|
|
|
