| T.R | Title | User | Personal Name
 | Date | Lines | 
|---|
| 192.1 | What seems to be the problem? | IOSG::TALLETT | Mit Schuh bish hi | Mon Mar 09 1992 09:10 | 12 | 
|  |     Hi there!
    
    	You don't say what your problem is. You say that the non-privved
    	account can't access the queue entries - so what? What error do
    	you get?
    
    	A1SUBMIT is for non-privved admin accounts to submit under MANAGER,
    	(with some security built in). You seem to be doing it the other way
    	around, so SUBMIT/USER= should be enough.
    
    Regards,
    Paul
 | 
| 192.2 | Queue protection | IOSG::SHOVE | Dave Shove -- REO-D/3C | Mon Mar 09 1992 15:54 | 7 | 
|  |     You can add an ACL to the queue, allowing Read access to that user.
    
    Or, if you want all user to be able to see entries on the queue, either
    add an ACL specifying user as * or change the queue protection from W:W
    to W:RW
    
    Dave.
 | 
| 192.3 | ACL aids | SNOC02::ZORBASSTUART | NULL Junior | Tue Mar 10 1992 01:42 | 34 | 
|  |     
    The idea was that some batch jobs (which do some processing and mail
    notification) could be executed by a non-privileged account and managed
    out of ALL-IN-1 housekeeping. 
    
    ALL-IN-1 housekeeping uses A1SUBMIT to submit the batch job, so I
    couldn't use other submit methods.
    
    If A1SUBMIT submits a job on behalf of a non-priviled user then that
    user cannot read his own entry on the queue, therefore it fails.
    
    A SHOW ENTRY/FULL from a privileged account does not show anything to
    why the non-priv User cannot see his own entry. (What magic does
    A1SUBMIT perform?)
    
    As .2 pointed out that putting an ACL on the queue now allows the user
    to read is own records (as well as everybody elses) and the job runs OK. 
    
    But, SMJACKET.COM then tries to create BLP in OA$LIB for a mail message
    of it's own, and of course my non-privileged user cannot write to
    OA$LIB. 
    
    Options:
    
    	- Don't use Housekeeping and write my own.
    	- Give the user privlege
    	- Use a priviled User (e.g. MANAGER)
    	- Modify the housekeeping procedures
    
    Cheers,
    
    Stuart
    
    	
 | 
| 192.4 | Roll your own | IOSG::TALLETT | Mit Schuh bish hi | Tue Mar 10 1992 08:36 | 61 | 
|  |     Hi There!
    
>    The idea was that some batch jobs (which do some processing and mail
>    notification) could be executed by a non-privileged account and managed
>    out of ALL-IN-1 housekeeping. 
>    
    	It would be kinda nice to know a few more details, what sort of
    	things the job is doing. Might help us suggest alternatives.
    
>    If A1SUBMIT submits a job on behalf of a non-priviled user then that
>    user cannot read his own entry on the queue, therefore it fails.
>    
    	You still don't say what fails, or what the actual error text is.
    	I still don't understand why the job WANTS to read its own queue
    	entry. Queue entries are read by the Job Controller, not by the
    	job.
    
>    A SHOW ENTRY/FULL from a privileged account does not show anything to
>    why the non-priv User cannot see his own entry. (What magic does
>    A1SUBMIT perform?)
>    
    	Well this works for me, I just tried it. The only magic in A1SUBMIT
    	is that it is installed with privileges, so after doing some
    	security checks, it does a privileged SEND_JBC, so you can specify
    	another /USER because it has privs. It should be the same as a
    	SUBMIT/USER in your case, because the submittor has privs already.
    
>    As .2 pointed out that putting an ACL on the queue now allows the user
>    to read is own records (as well as everybody elses) and the job runs OK. 
>    
    	Is *READING* all the records a big deal? He can't write the
    	records.
    
>    But, SMJACKET.COM then tries to create BLP in OA$LIB for a mail message
>    of it's own, and of course my non-privileged user cannot write to
>    OA$LIB. 
>    
    	I don't think SMJACKET was really designed to be used by
    	non-privved users. If you fix this problem (say by redefining
    	OA$LIB in the user process to be a searchlist) you may well run
    	into other problems.
    
>    Options:
>    
>    	- Don't use Housekeeping and write my own.
>    	- Give the user privlege
>    	- Use a priviled User (e.g. MANAGER)
>    	- Modify the housekeeping procedures
>    
    
    	I don't think you want to give the user privelege (and by the way,
    	write access to OA$LIB is equivalent to access to the MANAGER
    	account) and I personally wouldn't modify the housekeeping
    	procedures (but then I am bias! :-). Using a priv user (MANAGER)
    	is probably simplest, but I can't tell if that is a security
    	problem without knowing what you are trying to do. With the info
    	you have given so far, I would recommend writing your own, and not
    	trying to use housekeeping.
    
    Regards,
    Paul
 | 
| 192.5 |  | SNOC02::ZORBASSTUART | NULL Junior | Wed Mar 11 1992 04:48 | 24 | 
|  |     
    If, from the ALLIN1 account you:
    	
    	$ SUBMIT/USER=BLOGGS/AFTER=later....etc
    
    	$ A1SUBMIT/USER=BLOGGS/AFTER=later....etc
    
    Then from the BLOGGS (no privileges) account you:
    
    	$ SHOW ENTRY
    
    You will see two entries, however, the A1SUBMIT entry will display a 
    "no privilege".
    
    And as correctly pointed out this did not affect the running of the job
    as I first thought.
                     
    To get over the other lack of privlege problems (i.e creating the
    OA$LIB:.blp) I have given the account privileges until I come up with
    something better.
    
    	Cheers,
    
    	Stuart
 | 
| 192.6 | Don't understand whats going on here | IOSG::TALLETT | Mit Schuh bish hi | Wed Mar 11 1992 08:06 | 18 | 
|  |     
    >    If, from the ALLIN1 account you:
    >
    >        $ SUBMIT/USER=BLOGGS/AFTER=later....etc
    >
    >        $ A1SUBMIT/USER=BLOGGS/AFTER=later....etc
    >
    
    	Well that is an excellent test. I can't imagine what the
    	difference is here. Did you actually submit the same file?
    	Did show entry show ANYTHING of the A1SUBMIT job, (eg was
    	it complaining about the protection of the file you submitted
    	or about the queue entry itself).
    
    	Sorry we couldn't help on this one.
    
    Regards,
    Paul
 | 
| 192.7 | OA$ADMIN_DATA for .BLP? | IOSG::PYE | Graham - ALL-IN-1 Sorcerer's Apprentice | Wed Mar 11 1992 10:02 | 10 | 
|  |     Re .5
    
    Sorry if I missed this earlier, but why does the .BLP have to be in
    OA$LIB?
    
    Here's an idea for you: The ADMIN_DATA directory (OA$ADMIN_DATA
    logical) is writeable by an Administrator, and is specially intended
    for this sort of thing. Could you put your .BLP in there?
    
    Graham
 | 
| 192.8 | Job owner .NE. user to run under (?) | IOSG::SHOVE | Dave Shove -- REO-D/3C | Wed Mar 11 1992 16:42 | 9 | 
|  |     I think that it comes down to the owner of the queue entry (which
    neither SHOW QUEUE nor SHOW ENTRY show you).
    
    When a job is submitted to run under a different user, I think it's
    still owned by the submitting user. So the user under which it's to run
    may not be allowed to read it, depending on privs, queue protection
    etc.
    
    Dave.
 | 
| 192.9 | So why does SUBMIT/USER= work? | IOSG::TALLETT | Mit Schuh bish hi | Wed Mar 11 1992 16:45 | 1 | 
|  |     
 | 
| 192.10 | DCL is cleverer, I suppose | IOSG::SHOVE | Dave Shove -- REO-D/3C | Wed Mar 11 1992 16:58 | 7 | 
|  |     I suppose DCL SUBMIT sets the owner UIC to the same as that of the user
    under which it's to run.
    
    A1SUBMIT doesn't bother, because the poor simple programmer wasn't
    aware of this distinction at the time.
    
    Dave (aka Poor Simple Programmer)
 | 
| 192.11 | X marks the spot | SNOC02::ZORBASSTUART | NULL Junior | Wed Mar 11 1992 23:01 | 35 | 
|  |     
    re: the OA$LIB:.BLP - it's SMJACKET.COM which creates the BLP in case
    of a utility failure. OAUSER sounds like a nice place to generate BLPs.
    
    Any how,
    
    Here are the SHOW ENTRY details. 
    Job 935 was SUBMITted, job 936 was A1SUBMITted.
    
    From the ALLIN1 account:
    
  Jobname         Username     Entry  Blocks  Status
  -------         --------     -----  ------  ------
  X               ZORBAS         935          Holding until 12-MAR-1992 12:00
    On Batch queue SYS$BATCH
    Submitted 12-MAR-1992 08:43 /NOLOG /PRIORITY=100
    File: _$2$DUA103:[ALLIN1.SITE.LIB_BRITISH]X.COM;1
  X               ZORBAS         936          Holding until 12-MAR-1992 12:00
    On Batch queue SYS$BATCH
    Submitted 12-MAR-1992 08:43 /NOLOG /PRIORITY=100
    File: _$2$DUA103:[ALLIN1.SITE.LIB_BRITISH]X.COM;1
    
    
    From the ZORBAS account:
    
  Jobname         Username     Entry  Blocks  Status
  -------         --------     -----  ------  ------
  X               ZORBAS         935          Holding until 12-MAR-1992 12:00
    On Batch queue SYS$BATCH
    Submitted 12-MAR-1992 08:43 /NOLOG /PRIORITY=100
    File: _$2$DUA103:[ALLIN1.SITE.LIB_BRITISH]X.COM;1
  no privilege                   936          Holding until 12-MAR-1992 12:00
    
 | 
| 192.12 | You learn something new every day! | IOSG::TALLETT | Mit Schuh bish hi | Thu Mar 12 1992 08:23 | 31 | 
|  |     Hi there!
    
    	I did some experiments:
    
    	On VMS V5.4-
    
    sodom$ create x.com
    $       Job_num = p1
    $       uic = f$getqui ("display_entry", "UIC", Job_num)
    $       name = f$getqui ("display_entry", "username", Job_num)
    $       show sym uic
    $       show sym name
     Exit
    sodom$ submit oa$lib:sm_restart /after=12:00:00/user=allin1
    Job SM_RESTART (queue SYS$BATCH, entry 672) holding until 12-MAR-1992 12:00
    sodom$ @x 672
      UIC = "[ALLIN1]"
      NAME = "ALLIN1"
    sodom$ set comm oa$lib:a1submit
    sodom$ a1submit oa$lib:sm_restart /after=12:00:00/user=allin1
    Job SM_RESTART (queue SYS$BATCH, entry 673) holding until 12-MAR-1992 12:00
    sodom$ @x 673
      UIC = "[TALLETT]"
      NAME = "ALLIN1"
    
    	so my poor programmer friend is correct! I never even knew this
    	UIC existed! By the way, it appears to have changed on V5.5. There
    	A1SUBMIT seems to work like SUBMIT.
    
    Regards,
    Paul
 | 
| 192.13 | OA$SUBMIT privs | SAC::JOYCE | Leaving... and I don't give a XXXX | Thu Jun 04 1992 10:46 | 12 | 
|  |     
    A related question (I think)...
    
    The Security Police of one of our major customers are asking why the
    OA$SUBMIT image requires CMKRNL and SYSPRV. The documentation only
    mentions CMKRNL.  They want to know if SYSPRV is really needed and, if
    so, EXACTLY why (and if so, why ain't it in the documentation?).
    
    Thanks in advance.
    
    Andy
    
 | 
| 192.14 | To write to queues... | IOSG::PYE | Graham - ALL-IN-1 Sorcerer's Apprentice | Thu Jun 04 1992 11:02 | 13 | 
|  |     It got SYSPRV added to it in V3.0 so that we could be sure of always
    being able to write jobs onto the selected queue. I assume we must have
    had a problem report complaining about that.
    
    I wasn't aware that we normally listed all the privilege justifications
    in the doc. set, although we do justify them to the SQM police
    internally, before the product ships. I've posted the list of
    justifications in this (or previous versions of) notesfile in the past.
    
    Graham
    
    PS But hey, if you've got CMKRNL, You can easily get SYSPRV or anything
    else. Perhaps we shouldn't tell customers that....
 | 
| 192.15 | For the Administrator | UTRTSC::SCHOLLAERT | Sweden, here we come | Thu Jun 04 1992 11:14 | 19 | 
|  |     re.13
    
    SYSPRV was intruduced in 2.4. Looks like documantation was not
    updated.
    
    From the STARS:
    
    For the Administrator to be successful in cancelling housekeeping             
    procedures they would need the SYSPRV privilege.                              
                                                                                  
    Obviously this is not suitable to many installations, A1SUBMIT was            
    designed to solve problems such as this and needs to be installed             
    with SYSPRV privilege by the ALL-IN-1 startup procedure.                      
                                                                                  
    This problem has been fixed in ALL-IN-1 V2.4.      
    
    Regards,
    
    Jan
 |