[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference iosg::all-in-1_v30

Title:	OLD ALL-IN-1 (tm) Support Conference
Notice:	Closed - See Note 4331.l to move to IOSG::ALL-IN-1
Moderator:	IOSG::PYE

Created:	Thu Jan 30 1992
Last Modified:	Tue Jan 23 1996
Last Successful Update:	Fri Jun 06 1997
Number of topics:	4343
Total number of notes:	18308

176.0. "EM: Messages with duplicate file names" by KERNEL::LOAT (Bored....Bored....BORED!!!!) Thu Mar 05 1992 12:44

    
    A customer has hit on abit of a problem with his mail messages in
    ALL-IN-1 2.4 running under VMS 5.4-3
    
    He found a mail message which, when read, caused the users to be thrown
    out of ALL-IN-1, with a access violation. While investigating this, he
    found what looks like two mail messages with the same VMS filename in
    the same shared area!
    
    In the OAMTI_ERR.LOG, there is an error which indicates that it
    couldn't write the record, as there was a duplicate key. I assume that
    this is writing to the SDAF.
    
    Looking at a SH of both messages, the VMS filename is indeed the same.
    
    I was under the impression that this should never happen, with the
    introduction of the OA$FCV process. Is it possible if this process has
    died for some reason? The customer has worked around this, but now he's
    a bit worried about what caused this problem.
    
    Any ideas?
    
    Steve.

T.R Title User Personal
Name Date Lines

176.1 There are possible, but rare causes AIMTEC::PORTER_T Terry Porter, ALL-IN-1 Support, Atlanta CSC Thu Mar 05 1992 16:22 37

T.R	Title	User	Personal Name	Date	Lines
176.1	There are possible, but rare causes	AIMTEC::PORTER_T	Terry Porter, ALL-IN-1 Support, Atlanta CSC	`Thu Mar 05 1992 16:22`	37
	I've described this before, but here goes again ... The filename is generated based on date, time and a counter. The date determines the first 4 characters after the Z, the time (in minutes since midnight) plus the counter determin the last 4 characters of the filename. The counter is maintained in a lock and is incremented each time a filename is generated (it resets to 1 after 40,000). The OA$FCV image ensures that there is always a process with this lock (it takes out a null mode lock and then suspends itself), otherwise the lock and the counter value would be lost when all the user's exit ALL-IN-1. This mechanism has some small holes which can cause duplicate filenames ... - Creating 40,000 documents in one minuit (almost impossible) - If OA$FCV is not there and no-one is using ALL-IN-1 then creating a document loggining out of ALL-IN-1, logging back in and creating another document within 1 minute. It is unlikley that a user would do this but the Fetcher just might overnight on a reasonably fast CPU. - If the time is different on two nodes in a cluster such that 40,000 documents are created during that time difference plus one minute. - If OA$FCV is not running AND there is a time difference between 2 nodes in the cluster AND only one user is using ALL-IN-1, then creating two documents in the time difference plus one minute and logging out of ALL-IN-1 and back in inbetween the two document creations. If you think OA$FCV could have stopped then the most likely explination is that the fetcher ran twice overnight on two nodes in the cluster and the difference in time between the 2 nodes, plus 1 minute is longer than the time between the 2 fetcher runs. Unlikely but possible. HTH Terry

I've described this before, but here goes again ...

The filename is generated based on date, time and a counter. The date determines
the first 4 characters after the Z, the time (in minutes since midnight) plus
the counter determin the last 4 characters of the filename.

The counter is maintained in a lock and is incremented each time a filename
is generated (it resets to 1 after 40,000). The OA$FCV image ensures that there 
is always a process with this lock (it takes out a null mode lock and then 
suspends itself), otherwise the lock and the counter value would be lost when 
all the user's exit ALL-IN-1.

This mechanism has some small holes which can cause duplicate filenames ...

- Creating 40,000 documents in one minuit (almost impossible)

- If OA$FCV is not there and no-one is using ALL-IN-1 then creating a document
  loggining out of ALL-IN-1, logging back in and creating another document
  within 1 minute. It is unlikley that a user would do this but the Fetcher
  just might overnight on a reasonably fast CPU.

- If the time is different on two nodes in a cluster such that 40,000 documents
  are created during that time difference plus one minute.

- If OA$FCV is not running AND there is a time difference between 2 nodes in the
  cluster AND only one user is using ALL-IN-1, then creating two documents
  in the time difference plus one minute and logging out of ALL-IN-1 and back
  in inbetween the two document creations. 

If you think OA$FCV could have stopped then the most likely explination is that
the fetcher ran twice overnight on two nodes in the cluster and the difference
in time between the 2 nodes, plus 1 minute is longer than the time between the
2 fetcher runs. Unlikely but possible.

HTH

Terry