jhebert1
2 Bronze

Symm ESCON space reclamation

Jump to solution

Hi everyone. We have recently migrated all of our mainframe data off of one of our DMX1000s, and I am attempting to reclaim the space. There are 400 ESCON devices that I'd like to delete, and have never done this before. I have unmapped and deleted non-ESCON devices, and am using the same procedure, but the results are not encouraging.

The ESCON devices are not masked, but they are mapped to the various EA ports. When attempting to unmap the devices, I get an error message telling me:

Validating configuration changes..........................Failed.

    Error occurred while Validating

    The lowest ESCON channel number should be divisible by 0x10

The file I'm using for symconfigure is:

unmap dev 0000 from dir ALL:ALL;

I've also tried this with putting in the director number and port number (3a:0) and gotten the same results. Any thoughts/ideas??? Thanks in advance!

0 Kudos
1 Solution

Accepted Solutions
RobertDudley
3 Argentium

Re: Symm ESCON space reclamation

Jump to solution

You cannot unmap all devices from the escon channels, if you get the configuration to go through all the ESCON directors will drop into IMPL monitor mode DD.  If the directors go DD you cannot make any symcli changes.

Do it in this manner.

Unmap all devices with the exception of one control unit (one range of devices).

Create a new range of of mainframe devices with a new SSID it could be only one device

Map this range to all the escon directors as a new CU

go back and unmap the last full range of devices you left

some examples of creating and then mapping mainframe dev's.

create dev count=117, size=10017, emulation=ckd-3390, config=2-Way-Mir, mvs_ssid=E183;
create dev count=75, size=3339, emulation=ckd-3390, config=2-Way-Mir, mvs_ssid=E183;

mapping is a bit different as well this would be control unit 9  ( i was mapping both 3390-9's and 3390-3's into one CU that I created above) you map to one director and then copy to all the other directors you want the addressing on.

map dev 5e9:608 to dir 2a:0 starting base_address=900 mvs_ssid=E189;
map dev 61b:63a to dir 2a:0 starting base_address=920 mvs_ssid=E189;
map dev 5e9:608 to dir 3a:0 copying dir 2a:00;
map dev 5e9:608 to dir 14a:0 copying dir 2a:00;

0 Kudos
5 Replies
beetgj
1 Copper

Re: Symm ESCON space reclamation

Jump to solution

You are trying to unmap address X 00 on Escon which break IBM rules...escon addressing always start at 00 base address on every cu image. Below I extracted info from knowledgebase solution emc45750:

SymmWin at all Enginuity levels enforces the rule that the FIRST address in every CU must be divisible by hex 10.

  • For example, the FIRST address in every CU must be xx00 or xx10 or  xx20 or xxF0, etc. Otherwise, the Symmetrix bin file will not meet IBM addressing standards and the bin file therefore cannot be saved.
  • Note that gaps in the channel addressing is tolerated as long as the first address in the CU remains divisible by hex 10. For example xx00, xx02, xx03, xx51, etc.
  • This issue is most commonly encountered when deleting channel addresses via Solutions Enabler. You would need to delete all addresses from say xx00 to xx0F. This would leave the FIRST address in the CU as xx10 and this is acceptable to SymmWin as it meets the IBM addressing standards. Note that when deleting mainframe channel addresses (EA or EF) you will also need to simultaneously change the SSID value assigned in the bin file. Refer to the Solution Enabler Array Controls Product Guide on how to specify a new SSID within the symconfigure unmap command line.

Please also view knowledgebase solution or primus solution emc219945

RobertDudley
3 Argentium

Re: Symm ESCON space reclamation

Jump to solution

You cannot have a mainframe host channel without a device addressed on it.  You will have to leave at least one device addressed across all the escon adapters in your array.  SSID's also come into play it expects all device with the same SSID to be addressed the same.  You will have to do some SID work unless you want to leave an entire range behind.

0 Kudos
jhebert1
2 Bronze

Re: Symm ESCON space reclamation

Jump to solution

First of all, THANKS for the reply.

What if I unmapped all ESCON devices? I should go back in and map one device back to two controllers, right?

e.g.

map dev 0000 to dir 3a:0, starting base_address=A20;

map dev 0000 to dir 14a:0, copying dir 3a:0;

Would this work to re-map device 0000? My experience is with FBA, not ESCON, so any assistance would be great!

BTW, I do have a call in with EMC, but am waiting on the appropriate resources to be available.

0 Kudos
RobertDudley
3 Argentium

Re: Symm ESCON space reclamation

Jump to solution

You cannot unmap all devices from the escon channels, if you get the configuration to go through all the ESCON directors will drop into IMPL monitor mode DD.  If the directors go DD you cannot make any symcli changes.

Do it in this manner.

Unmap all devices with the exception of one control unit (one range of devices).

Create a new range of of mainframe devices with a new SSID it could be only one device

Map this range to all the escon directors as a new CU

go back and unmap the last full range of devices you left

some examples of creating and then mapping mainframe dev's.

create dev count=117, size=10017, emulation=ckd-3390, config=2-Way-Mir, mvs_ssid=E183;
create dev count=75, size=3339, emulation=ckd-3390, config=2-Way-Mir, mvs_ssid=E183;

mapping is a bit different as well this would be control unit 9  ( i was mapping both 3390-9's and 3390-3's into one CU that I created above) you map to one director and then copy to all the other directors you want the addressing on.

map dev 5e9:608 to dir 2a:0 starting base_address=900 mvs_ssid=E189;
map dev 61b:63a to dir 2a:0 starting base_address=920 mvs_ssid=E189;
map dev 5e9:608 to dir 3a:0 copying dir 2a:00;
map dev 5e9:608 to dir 14a:0 copying dir 2a:00;

0 Kudos
P7EXKX
1 Copper

Re: Symm ESCON space reclamation

Jump to solution

We have the same problem where we want to get rid of all storage formerly addressed to a Mainframe never existing.

So I could remove the whole pools and the data devs and tdevs in it except of a few.

What do you mean with "SID work" and what is this in detail ?

I want to use these devs with a ckd-3390 format for oen systems with FBA.

0 Kudos