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

Conference vaxaxp::vmsnotes

Title:	VAX and Alpha VMS
Notice:	This is a new VMSnotes, please read note 2.1
Moderator:	VAXAXP::BERNARDO

Created:	Wed Jan 22 1997
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	703
Total number of notes:	3722

695.0. "relative file backup problem" by VAXSPO::ANDRE () Thu Jun 05 1997 11:22

	I have a customer with two 8400 5/440 that perform the same function.

   One is the primary and other is the secondary .

	Every night they back up certain database files from the primary to

   a temporary disk , from the temporary disk to a TZ88 , and from the TZ88

   to the secondary .

	After this they have a software that synchronizes the contents of

   the two systems .

	But there is one specific file that fails in this synchronization .

	The status indicates that some records are different .

	It's a relative file with 1500015 blocks .

	When they don't use the temporary disk , in other words , when they

   do a direct backup to the TZ88 , the restore operation works fine .

	The disks are the same ( RAID array of RZ29s controlled by HSJ52 ) ,

   and they use sample backup commands .

	The cluster size of the database files disk is 33 . The temporary 

   disk is 17 . Does it matter ??

	When restoring the backup they're using /OVERLAY .

	Does anyone know any problem like this ?

	Any help will be appreciated .



			Andre'

T.R	Title	User	Personal Name	Date	Lines
695.1	Various Questions...	XDELTA::HOFFMAN	Steve, OpenVMS Engineering	`Thu Jun 05 1997 11:55`	25
	Enable highwater marking on the file. (This will have no adverse performance affects, assuming they are not regularly writing new records well past the end of the current highest record -- that they are not greatly extending the files.) See if this cleans up the comparision operation. Exactly what BACKUP commands are used, and are there actually any differences between the source and target files visible via tools such as DIFFERENCES? Are these folks using /IGNORE=INTERLOCK, or are the data files open for write access by running applications? What OpenVMS version? Why are they using this particular scheme, and not a VMScluster, or a VMScluster with the MDF/BRS package? Would they be amenable to alternative file transfer schemes, or are they set on this approach? Would they be interested in using DEC RTR or similar distributed transaction-processing packages? What is the physical distance between the two systems?
695.2	* more info *	VAXSPO::ANDRE		`Fri Jun 06 1997 09:21`	54
	Steve , The backup command is a "$ BACKUP/LOG LOGICAL:. TEMPLOGICAL:.". To restore they use the "/OVERLAY" qualifier . I couldn't compare the two files because the only moment that they're identical is when they stop the system to restore the files and synchronize the whole thing . And this moment is always in the middle of the night . What I did is to check the file structure and size . These attributes are ok . They're not using /IGNORE=INTER because when they do the backup , they always stop the application that access the files . OpenVMS version is 6.2-1H3 . They're not using cluster because they have a kind of application developed by GTECH corporation that mirrors all the systems . So when a process start in one machine , this software creates the same process with the same characteristics in the other machines . This customer is one of our biggests sites in our country . They have two sites ( one primary and one backup ) with a lot of duplicated systems (9 Alphastations 255 5/300 , 4 4100 5/400 , 4 8400 5/440 ) and they control all kinds of lotteries and games in the whole country . I don't think that they can change anything on this scheme . It's a kind of " black box " even for us . They're very confidentially about the way that the application works . The two systems in question are in the same site . thanks , Andre'
695.3	More Questions, No Answers...	XDELTA::HOFFMAN	Steve, OpenVMS Engineering	`Fri Jun 06 1997 10:43`	30
	: After this they have a software that synchronizes the contents of : the two systems . : But there is one specific file that fails in this synchronization . : The status indicates that some records are different . Please provide more information on the synchronization software -- the BACKUP command shown will not perform this comparision -- and provide more information on the records that are different. (Please also provide the BACKUP command used to go to tape, and then to restore the tape on the target disk.) I will assume you have already contacted the support folks from GTECH, and have gotten them involved in this, as they are the folks that may (will?) be able to provide us with more details on the operation of the internals of the package. And they may have seen this problem before. : It's a relative file with 1500015 blocks . Please provide us with a DIRECTORY/FULL on the source and "corrupt" files, and please provide us with a SHOW DEVICE/FULL on the source and target disks. : The cluster size of the database files disk is 33 . The temporary : disk is 17 . Does it matter ?? I'd try making this setting consistent. (We do not know enough about the whole process to determine if this setting is relevent -- I'd assume not, but I have no idea what sort of comparision is being performed.)