Hercules Initialization Processing

[halfmeg]

Have been tinkering with the TK3 update before finalizing and encountered a odd situation, to me anyway.

I added a string of 3390s at E00-EFF, 256 in all.

When Hercules is started without any devices in this range and an attempt is made to 'attach' one after startup but before IPL of the OS (MVS 3.8j), it isn't available to the OS after IPL.

    - STC  169  IEF244I MOUNT E10 - UNABLE TO ALLOCATE 1 UNIT(S)
    -         AT LEAST 1 OFFLINE UNIT(S) NEEDED.
    - STC  169  IEF489I MOUNT - 1 UNIT(S) NEEDED FOR IEFRDER
    - STC  169  IEF247I MOUNT - E10 NOT ACCESSIBLE
     *STC  169 *01 IEF238D MOUNT - REPLY DEVICE NAME OR 'CANCEL'.
 

                $HASP309    INIT  1 INACTIVE ******** C=A
      JOB   30  $HASP150 JOB3     ON PRINTER1    21,211 LINES
  00            $HASP160 PRINTER1 INACTIVE - CLASS=AP
      JOB   30  $HASP250 JOB3     IS PURGED
    -           d u,,,e00,4
    -           v e00,online
                IEE025I UNIT E00 HAS NO LOGICAL  PATHS


When a entry for E00 or E10 is in the Hercules config before start of Hercules, then the pack is available to the guest OS without any problems.

Is there an explaination or solution for the above other than having a 'dummy' entry in the Hercules config for each channel ?

Phil

[kerravon86]

--- In hercules-390@..., "halfmeg" <opplr@...> wrote:
>
> When a entry for E00 or E10 is in the Hercules config
> before start of Hercules, then the pack is available
> to the guest OS without any problems.

Sounds like Hercules has some "max device number".

Does defining a dummy FCC (etc) allow it to work
for 000-FCB?

BFN.  Paul.

[Kevin Leonard-2]

> v e00,online
> IEE025I UNIT E00 HAS NO LOGICAL  PATHS
>
> When a entry for E00 or E10 is in the Hercules config
> before start of Hercules, then the pack is available
> to the guest OS without any problems.
>
> Is there an explaination or solution for the above
> other than having a 'dummy' entry in the Hercules
> config for each channel ?

If there are no devices at all, ready or not ready,
on a sysgened channel, MVS puts the channel offline at
startup.  You could vary channel E online, then issue
the VARY ONLINE for E00.

It may be less confusing for the novice user to just
have a dummy device on each channel.

--
Kevin

[Tony Harminc]

2009/10/7 Kevin Leonard <kleonard_list@...>:

>
>
>> v e00,online
>> IEE025I UNIT E00 HAS NO LOGICAL  PATHS
>>
>> When a entry for E00 or E10 is in the Hercules config
>> before start of Hercules, then the pack is available
>> to the guest OS without any problems.
>>
>> Is there an explaination or solution for the above
>> other than having a 'dummy' entry in the Hercules
>> config for each channel ?
>
> If there are no devices at all, ready or not ready,
> on a sysgened channel, MVS puts the channel offline at
> startup.  You could vary channel E online, then issue
> the VARY ONLINE for E00.

I was going to say about the same thing, but I think the problem as
described is more subtle. Phil says:

>>When Hercules is started without any devices in this range and an attempt is made to 'attach' one after startup but before IPL of the OS (MVS 3.8j), it isn't available to the OS after IPL.

This sounds as though Hercules is doing something different if the
device is ATTACHed, even before the OS has had any opportunity to test
for its existence. Does Herc really have a notion of channels? I
understood everything was just devices.

Tony H.

[Kevin Leonard-2]

> > If there are no devices at all, ready or not ready,
> > on a sysgened channel, MVS puts the channel offline at
> > startup.  You could vary channel E online, then issue
> > the VARY ONLINE for E00.
>
> I was going to say about the same thing, but I think the
> problem as described is more subtle. Phil says:
>
> >> When Hercules is started without any devices in this
> >> range and an attempt is made to 'attach' one after
> >> startup but before IPL of the OS (MVS 3.8j), it isn't
> >> available to the OS after IPL.
>
> This sounds as though Hercules is doing something different
> if the device is ATTACHed, even before the OS has had any
> opportunity to test for its existence.

Prior to about mid-2007, there was a problem where Hercules
was always making a CRW pending when a device was added by an
attach command, even if the machine was currently running in
S/370 mode.  This would cause problems later, when device
processing would get errors because Hercules expected the
pending CRW to have been handled.  S/370 operating systems
that don't know about CRWs didn't deal with this consistently
well, with the result that it was difficult or impossible to
attach devices after IPL and have them recognized by the
software.  That's fixed in all modern copies of Hercules.

--
Kevin

[halfmeg]

> Kevin Leonard wrote:

> If there are no devices at all, ready or not ready,
> on a sysgened channel, MVS puts the channel offline at
> startup.  You could vary channel E online, then issue
> the VARY ONLINE for E00.

This works.

Attach device on previously unoccupied channel in hercules console.

Switch to MVS console.

Enter "v ch(x),online"
Enter "v yyy,online" or enter a mount command.

> It may be less confusing for the novice user to just
> have a dummy device on each channel.

Thanks,

Phil

[halfmeg]

> Tony Harminc wrote:

><snip>

> I was going to say about the same thing, but I think the problem as
> described is more subtle. Phil says:

> > When Hercules is started without any devices in this range and an
> > attempt is made to 'attach' one after startup but before IPL of
> > the OS (MVS 3.8j), it isn't available to the OS after IPL.

> This sounds as though Hercules is doing something different if the
> device is ATTACHed, even before the OS has had any opportunity to test
> for its existence. Does Herc really have a notion of channels? I
> understood everything was just devices.

Although I think I stated it correctly as it occured, I could have been wrong.  Varying the channel online before vary of the device online cures the situtation.

Phil