VMS Retirement meeting for 13 July 1999

This meeting was attended by Darien Wood, Nick Hadley,  Wyatt Merritt, Dave Fagan, Dane Skow, Alan Jonckheere,
Jean Rhoades, Mike Diesburg, and Neal Schmidt.  The meeting had the following topics (in literally) no particular order:
Status of VMS backups of tmp and prj$root disks
Applications not ported to UNIX
Stages of  VMS retirement
VMS Retention Requests
What to do next Tuesday

 

Status of VMS backups of tmp and prj$root disks

The FNALD0 sets of backups have been completed.  We are currently "switching"  the tape labels since the
verified tapes were the ones that originally were slated to be given to d0.  Andrew Brandt and Darien requested that
these tapes be tested on the UNIX system to determine that it is possible to retrieve data from them.  Jean Rhoades
and Neal Schmidt were going to go and get a couple of tapes to test the VMSbackup utility [this was done and Jean
is testing her procedures].

The TNG backups are in the first stage of making backup tapes.

The assumption that there "exist other copies of important data (as that was the data area manager's responsibility) and that these are emergency backups, only" was restated and reaffirmed by the group.
 
 

Applications not ported to UNIX

An attempt was made to determine which applications were "not going to be ported to UNIX" and which "cluster" those
applications were run from.
 
 
FNALD0 TNG D0FS D0SFT D0IT
Trigsim DECnet Phase V router
Pixcie Pixie
Pickevents Client Pickevents Client Pickevents Server
Luminosity Luminosity
RDB RDB
Fatman Client Fatman Client Fatman Server

Although there were concerns expressed, it was general agreement that "no applications were only available on the
TNG cluster".  There were applications that "could only be run a VAX" so if only one of the two workstation clusters
remained it had to be FNALD0.

Alan is the only known user of D0SFT and tasked with migrating the distribution functions off the machine. The
retirement date for D0SFT was September 1, 1999.
 
 

Stages of  VMS retirement

It was determined that there were three stages of retiring VMS machines:
 

Stage 1 (Desktops)

Stage 1 was made up of removing the need to support VMS desktops.  This included the following:
Removing the data areas of desktop machines so that workstations no longer had to serve disks
Replace the screens so that users no longer needed the desktop VMS machines
Retirement dates for data areas have been posted (and are reprised below) and area managers should move or backup important information.  Backups will be made to protect forgotten or orphaned data.

On December 1, 1999 all VMS (desktop ?) machines at D0 will be retired, regardless of clusters. The cluster cores, located in Feynman, will continue to provide the identified residual functions required. By December 1, all purchases necessary to replace the VMS workstations need to be complete and the support arrangements made. The approved D0 Run II desktop is a PC and guidance can be found <where?>.

The following is the current environment for each cluster.
 
TNG
FNALD0
D0IT
Total nodes
25
 37
12
Boot nodes
1
5
2
Server nodes
2
2
0
Fermi owned workstations
2
5
6
University owned workstation
20
19
4
Down or dead machines
4
20

On Tuesday, July 20, the DOIT cluster will be retired. Screens that have been identified as needing to continue will be served either by FNALD0 nodes or by replacement with PCs.  This may take a little longer to implement.

Stage 2 (Core functions)

Stage 2 was made up of consolidating the remaining machines to handle the core functions.

It was decided that for FNALD0 the core functions could be consolidated into D0GS01, D0GS06 and D0GSLB.
There was some discussion that the core functionality was really just D0GS01 and D0GSLB.  A determination
of the remaining disks would have to be made but it was desired to keep the 6 home area disks, the two
scratch disks (for emergencies),  the d0 and LB library disks and the Fermi d0 products disks.

Stage 3 (indefinite Run I)

Stage 3 was designing a VMS environment to handle any future requirements.

A number of possible designs were discussed.  One was to combine the "remains" of the FNALD0, TNG and DOFS
clusters into one remaining cluster.  This would involve replacing the Client Server relationship of D0FS to the other
two clusters and nobody was sure how to do it (or even if it was possible to do).

The alternative was to consolidate D0FS, TNG and FNALD0 into as small as possible clusters and "living with the results"
There was general agreement that this solution was feasible and a likely requirement for any alternative solution.
 
 

VMS Retention Requests

It was determined that the following deadlines would be used to determine if a user
should continue their analysis or convert.
Analyses using FNALD0 resources need to be submitted for publication
by August 1st, or they should be ported to Unix systems.

Analyses using TNG resources need to be submitted for publication
by Sept. 1st, or they should be ported to Unix systems.

Analyses submitted for publication before the deadline may keep necessary resources until they have been published.

There was a request by Nick Hadley that "somebody" (Darien Wood or designee) needed to build a list of "successful users that have already converted to UNIX".
 

The following requests for for disk extensions were discussed:

>tmp$root333    STT simulator      > 1st Sept          Meena
It was decided that Meena needs to convert this.

>tmp$root358     Di-Lepton mSUGRA   Until PRL accepted  Genik
>tmp$root320     analysis           or D0TNG shutdown
>tmp$root353
These were approved until Sept 1st or the user would have to convert

>tmp$root336
This was approved until Sept 1st or the user would have to convert

>prj$root317     W+jets color       Until publication   Varelos
>tmp$root320     coherence
These were approved until Sept 1st or the user would have to convert

>Raja1           Energy calibration  Forever            Raja
>Raja3           code
>Raja4
This was NOT approved.  The backups were determined to be sufficient

>tmp$root346     QCDWZ ntuple maker  Until port done    Gerber
This was approved until Sept. 1st

>prj$root313     Wpt, Zpt, alpha-2   May 2000           Gerber
The Wpt project needs to convert to UNIX now
The Zpt was approved until Sept 1st (or needs to be converted)
The alpha-2 was unlikely to make the Sept 1st deadline so it should be converted, now.

>tmp$root316     some analysis       for thr summer     Casey
This was approved until Sept 1st or the user would have to convert

>prj$root352     trigger studies     1 month            Hirosky
>prj$root345     QCD Jets PRD        Month or so        Hirosky
These were approved until Sept 1st or the user would have to convert

>tmp$root329     Pomeron exchange    September?         Hirosky
The Pomeron exchange needs to be converted since it would probably not
 make the Sept 1st. deadline.

>prj$root310     QCD                 > summer           Babukhadia
>prj$root327     QCD
>prj$root310     QCD analysis        indefinite         Nang
>prj$root327     QCD analysis        indefinite         Nang
These were approved until Sept 1st or the user would have to convert

>prj$root325     upg simulator       indefinite         Nang
This needs to be converted now.

>dair01$dka600   WZ cross section    until publication  Gomez
This was approved until Sept 1st or the user would have to convert

>prj$root307     "files accessed"    indefinite         Gerber
Not approved unless more information was provided

>tmp$root332     "finish a few jobs" indefinite         Landsberg
Approved until Aug 1st or a better explanation was provided

>NEW_PHENOM_4    SUSY R-parity       1 week             Qian
This was approved until Sept 1st or the user would have to convert
 
 

The following requests for extending workstations were discussed:

>d0fm18          DOIT node          "For the time being"Grinstein ditto
                                    until IP assigend  Grinstein Reasonable
This was approved until PC replaced it

>d0fm12          DOIT node           until replaced     Fuess     For Chinese collab.
>d0fm22          DOIT node           until replaced     Fuess     For Chinese collab.
These were approved until Stu Fuess can replace them

>d0ru01          DOIT node           indefinite         Adams
This was approved until Mike Diesburg can replace it

>d0fm26          DOIT                until PC arrives   Cummings
This was approved until PC arrives

>d0isu2          fnald0                indefinite         Mauritz
Approved until Mike Diesburg can replace it

>d0az07          DOIT                until PC arrives   Johns
Approved until PC arrives

>d0cd03          xfer Tau stuff,mail 2 weeks            Qz. Li
Approved until Aug 1st

>d0br03          DOIT node          "For the time being"
>d0hs03          EWS Station!!       indefinite         Gomez
>d0fm11          Support d0hs03      indefinite         Gomez
>d0ku01          fnald0                   ?                  Baringer
It was decided that Mike Diesburg and Dane Skow would replace these

>daumi1          TNG              indefinite         Turcot
>dana01          TNG              indefinite         Martin
TNG workstations are not scheduled for decommissioning until Sept 1st
 
 

What to do next Tuesday

Because the Tuesday, 13-Jul-1999, downtime for fnald0 and D0IT was postponed, there was a
discussion to determine what to do the following Tuesday.  It was decided that the D0IT retirement
for nodes that "nobody requested an extension on" could continue.  In addition, next Tuesday,
20-Jul-1999, we would shutdown all of the "eligible" fnald0 workstations.  These were the ones
that we have not heard any complaints or requests for extensions.  Following the downtime, work
would precede to replace the remaining D0IT machines with PC's (or newly available FNALD0
machines, if necessary).  However,  D0IT should be considered "retired" on the 20-Jul-1999 date.