Please login or register. June 23, 2017, 10:18:34 AM

Author Topic: Apple Network Server 500/132 & AIX 4.1.4  (Read 58591 times)

0 Members and 1 Guest are viewing this topic.

Leografix

  • Full Member
  • ***
  • Posts: 31
  • Karma: +0/-0
    • BlackMac.ch
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #20 on: October 08, 2008, 06:35:52 AM »
With installation media in the drive try:

installp -d /dev/cd0 -L | grep device.pci.2310

Ok, I double-checked my typing (allthough I did not find the "|" key on my swiss german Apple keyboard, I had to copy paste it). When using the command

installp -d /dev/cd0 -L | grep device.pci.2310

while AIX 4.1.5 install CD is in the drive it throws me an 0511-123 The volume on /dev/cd0 is not in backup format followed by a 0503-003 installp: the specified device /dev/cd0 is not a valid device or file error.

As I copied all installation files to my hard drive I re-tried this with the following command:

installp -d /usr/sys/inst.images -L | grep device.pci.2310

After hitting ENTER no error message appears but also no feedback at all. The block cursor just jumps to the next line and blinks. Has it installed something?

J

Leografix

  • Full Member
  • ***
  • Posts: 31
  • Karma: +0/-0
    • BlackMac.ch
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #21 on: October 08, 2008, 08:31:24 AM »
lscfg -vl ent0

Thanks a lot for this, I?m beginning to understand some command line basics by now. I determined the MAC adresses of the onboard ethernet of the ANS and the MAC adress of the SMC NIcard (a DEC NIC clone I suppose). The MAC adress of the 100 MBit Apple NIC isn?t recognized (who wonders...).

I?m not sure if You understood my DHCP question. I already noticed that AIX can act as a DHCP server (although I still don?t know what to configure where) but this is not what I?m aiming at.
I?m using a D-Link router which acts as a DHCP server providing static IP adresses to all of my machines in accordance to the MAC adress of their NI. I don?t want to set up my ANS as DHCP routing machine, it just has to "get" it?s static IP adresses from the router while every NI in the ANS gets it?s own IP adress (a maximum of 7 possible IP adresses as the ANS has one onboard ethernet port and up to six possible PCI NI ports).

I already reserved my wanted IP adresses within the D-Link router and bound them to the MAC adresses of the ANS. Sorry for being not too exact with my question: is AIX able to "get" the IP adresses provided by the D-Link -> dynamically <- or do I have to configure AIX -> manually <- to self-assign this IP adress? For example NeXTstep 3.3 isn?t able to get IP adresses dynamically (unless you are installing an additional DHCP package NOT from NeXT). If You are not using this package You have to edit Your network settings by hand "reserving" the wanted IP adress inside the NeXTstep OS AND (!) within the used D-Link router.

J

Michael

  • Administrator
  • Hero Member
  • *****
  • Posts: 1039
  • Karma: +0/-0
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #22 on: October 08, 2008, 03:04:31 PM »
Re: DHCP.

If you have configured your DHCP server to reply with a specific IP address to a specific MAC address then the command:

# smitty usedhcp

will take you into the AIX smit dialog to select the interface (en0, en1, etc) that you want configured via the DHCP client.

Re: installp ... -L - this only lists (L) the software found on installation media. As the first command gave an error, I must be making an error myself. For syntax you might try: installp -? to get the syntax message.

lslpp -L >x; grep 2310 x (rather than the pipe | character) to see if the devices.pci.2310* are installed.

Question: if you run cfgmgr is it still complaining about missing drivers?

Leografix

  • Full Member
  • ***
  • Posts: 31
  • Karma: +0/-0
    • BlackMac.ch
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #23 on: October 10, 2008, 01:31:27 PM »
When running cfgmgr it throws 0514-040 Error initializing a device into the kernel message (Method Error (/usr/lib/methods/cfgde)).

When running lslpp -L >x; grep 2310 x simply nothing happens, just a line break (the "#" now missing) and a blinking block cursor. So I don?t think it installed anything.

In the meantime I was able to "configure" TCP/IP at least a bit. Might be a small step for You but for me it is indeed a very large one! The AIX interfaces are still very confusing to me. So for now a 10MBps line is up on et1 and et0 and I was able to ping these from Mac OS X so this first step is the basis for all following =)

J
« Last Edit: October 10, 2008, 02:20:33 PM by Leografix »

Michael

  • Administrator
  • Hero Member
  • *****
  • Posts: 1039
  • Karma: +0/-0
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #24 on: October 10, 2008, 05:45:25 PM »
et0 and et1!

That is unusual. Seems Apple is using IEEE 802.3 rather than "Xerox Park" protocol.

The OSI layer two (or LAN) devices are entX and the OSI layer three devices are enX and etX.

regarding lslpp -L >x

what does wc x say. Maybe there is no software for you device on the CDROM.

I will try to figure which device is the method cfgde (cfgent is for configuring ent, or ethernet interfaces iirc. So I am looking for a device best described by 'de'.

Leografix

  • Full Member
  • ***
  • Posts: 31
  • Karma: +0/-0
    • BlackMac.ch
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #25 on: October 11, 2008, 05:31:57 AM »
et0 and et1!

That is unusual. Seems Apple is using IEEE 802.3 rather than "Xerox Park" protocol.

The OSI layer two (or LAN) devices are entX and the OSI layer three devices are enX and etX.

Well, I'm not sure about this but it seems to me that any network interface on the ANS runs two protocols at the same time within every single interface adapter. When calling the communication devices from SMIT every interface offers a "et" and an additional "802.3" so it's up to You to choose one.

I've been fiddeling around with that box yesterday quite some time and I discovered several things that appear unusual to me. Regarding networking it took me some time to figure out that one single TCP/IP stack is handling ALL network interfaces at the same time - before I thought there is a single stack for every single interface as the ANS can handle up to seven interfaces at the same time.

Another strange thing which I have to resolve: the ANS utilizes a level 2 cache chip which is - to my surprise - not "configured" after a clean BOS install. I know that AIX notes every single chip and device so I thought as long as this is a "standard" chip from  Apple their version of AIX should have been "installing" it. I didn't notice it from within SMIT but from the early startup diagnostics (didn't know until yesterday that turning the key switch in to service position before boot up it evokes this service diagnostics). Compared to other Macs this machine really lacks in-deep documentation in very essential points.

I also noticed that the ANS can handle a Quantum 40 GB DLT tape drive for backups but - instead of the built-in DDS2 tape drive - it can not rewind the tape for example in  order to boot of that DLT tape and restore the system. When booting off the installation CD the tape can at least be used for OS and system restoration. At any other point the DLT seems to work fine. The DDS2 can boot and install / restore but I don't know how to activate the hardware compression (must be software-driven as this Apple tape DDS doesn't feature jumpers for this option). 4 GB of tape space is not as much as I may need in the future. Thanks to the DLT it uses hardware compression so I'm having the option of 80 GB per tape.

Another odd thing about the ANS: Apple states the tray drives as "automatically terminated and SCSI ID given". While the drives really do get their correct SCSI ID dependig on the drive tray they are in the SCSI chain is not terminated or at least not terminated correctly. You still have to use jupers for that one meaning that several Seagate drives can not be used at the end of the chain because they lack termination by jumper setting.

After all my ANS seems to be very close to an IBM 43p machine - a Pro at least told me that  so I hope this can solve some compatibility questions a bit better.

J
« Last Edit: October 11, 2008, 05:45:46 AM by Leografix »

Michael

  • Administrator
  • Hero Member
  • *****
  • Posts: 1039
  • Karma: +0/-0
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #26 on: October 11, 2008, 08:27:57 AM »
AIX uses one IP stack for IPv4, all interfaces, and in the newer versions (starting with AIX 4.3.0) an additional stack for IPv6.

Focusing on IPv4 and AIX 4.1: each ethernet adapter runs at one speed (10 or 100) and either half or full duplex, depending on driver support.

The configuration of the adapter can be read by: lsattr -El entX
The actual status of an active adapter can be read by: entstat -d entX (output similiar to netstat -m, but not exactly the same)

Each ether adapter supports two IP protocols: what I call "Xerox Park" - iirc the original Ethernet LAN (layer 2) description; and IEEE 802.3 (also know as dot 3, token ring, e.g. is dot 5, or 802.5, etc., etc..). In AIX IP interface terms these interfaces are, respectively: { adapter, Xerox, dot 3 } :: { entX, enX, etX }

IP interfaces can be configured via ODM settings via:
chdev -l enX -a netaddr=10.168.1.1 -a netmask=255.255.255.0 -a state=up
or directly via ifconfig
ifconfig enX 10.168.1.1 netmask=255.255.255.0 up

To see if an interface is configured via ODM, or what it should ODM think it should be:
lsattr -El enX

To bring an interface down/up without destroying any configuration information:
rmdev -l enX or ifconfig enX detach ## may also clear some routing information
Online: mkdev -l enX

To restore default routing, as far as interfaces permit:
mkdev -l inet0

These commands are what smit is executing for you. Review $HOME/smit.script to see what commands smit has performed. $HOME/smit.log has the output from these commands.

John Peck

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 46
  • Karma: +0/-0
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #27 on: October 12, 2008, 02:08:54 AM »
It is not usual to see the level 2 cache configured as a device - proc0,1,... for the processors is as far as it usually goes,
the cache is being used seamlessly.

Normally tape drives are configured with hardware compression enabled.  That can be disabled with "smit chgtpe" as a rule.
It's also possible to use the dot number extensions to the device file to call on attributes of a tape drive - although not compression as such, just density (i.e. different versions of tape formats as in DDS2, 3, 4 etc).  From "man rmt":

Special File Name Rewind-on-Close Retension-on-Open Bytes per Inch 
/dev/rmt* Yes No Density setting #1                                 
/dev/rmt*.1 No No Density setting #1                               
/dev/rmt*.2 Yes Yes Density setting #1                             
/dev/rmt*.3 No Yes Density setting #1                               
/dev/rmt*.4 Yes No Density setting #2                               
/dev/rmt*.5 No No Density setting #2                               
/dev/rmt*.6 Yes Yes Density setting #2                             
/dev/rmt*.7 No Yes Density setting #2   

 1. The values of density setting #1 and density setting #2 come from tape     
    drive attributes that can be set using SMIT. Typically density setting #1 
    is set to the highest possible density for the tape drive while density   
    setting #2 is set to a lower density. However, density settings are not   
    required to follow this pattern.                                           
 2. The density value (bytes per inch) is ignored when using a magnetic tape   
    device that does not support multiple densities. For tape drives that do   
    support multiple densities, the density value only applies when writing to
    the tape. When reading, the drive defaults to the density at which the tape
    is written.                                                               
 3. Most tape drives use 512-byte block size. The 8mm tape drive uses a minimum
    block size of 1024 bytes. Using SMIT to lower the block size, will waste   
    space.                                                                                               


Another one to look at is the "tctl" command, which does rewinding for you.

In the actual 43P machine the internal SCSI cable is a loop from mother board to devices
and back again - each device has to have jumper set SCSI id and the termination is
in the motherboard.  If you connect inappropriate devices (wrong SCSI type), that will
give SCSI termination errors - only a select band of devices is supported and will work !

Anyone would think my knuckles drag on the ground now  :D

Leografix

  • Full Member
  • ***
  • Posts: 31
  • Karma: +0/-0
    • BlackMac.ch
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #28 on: October 13, 2008, 10:05:53 AM »
It is not usual to see the level 2 cache configured as a device - proc0,1,... for the processors is as far as it usually goes,
the cache is being used seamlessly.

Hmm... When running "diag -a" it mourns about a new resource that -> may <- require software installation:

- L2cache0     00-L0     L2 Cache

Within the ANS the Level 2 cache is a single additional chip which is placed on the mainboard and not the processor board. Removing it is no option as I know what can happen from other Macs.

J

John Peck

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 46
  • Karma: +0/-0
Re: Apple Network Server 500/132 & AIX 4.1.4
« Reply #29 on: October 13, 2008, 11:29:15 AM »

Actually, as Simon Cowel says on occasion, I got that wrong. 
L2cache0 I see as "available" on a proper machine with such cache using "lsdev -C".
If your box genuinely requires software to be installed for it, then I would expect
the device not to be "available" but "defined".  If it is "defined", it's not in use,
forget about it.  If it's available, it's in use and will make no difference whatever
I expect.