We have test and production accounts on the same system, which is to say that they reside in different paths on HP-UX.
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
Page 1 / 1
We have test and production accounts on the same system, which is to say that they reside in different paths on HP-UX.
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
Since the number of parts are the same and they "map" to the same numbers, yes. Just make sure that any indexes are also copied and that the copied part file is pointed to the TEST index location and not the production index location after copying. You don't want updates to the "copied" part files updating the production indexes. Always double check to make sure whatever pointers you are using are interfacing with your copies and not the production part files.
Once copied, use SET.INDEX, VERIFY.DF and LIST.DF to make sure everything looks good. If needed, you may have to use REBUILD.DF but I don't think that will be needed in your case.
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
SET.INDEX, LIST.DF, VERIFY.DF should show you the parts and indexes for the distributed file.
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
Not entirely sure what you mean by "stub files". If you mean the file that "defines" the distributed file, that is a type 27 file that lives under the account that it is created in. This contains no actual data, it just tells the system what part files define this distributed file (and the algorithm to use).
d) Where can I find the latest, on-line documentation about this?
Chapter 6 of the UniVerse system description.
Rocket Software Documentation
------------------------------
Ryan Ladd
------------------------------
We have test and production accounts on the same system, which is to say that they reside in different paths on HP-UX.
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
To clarify, are you copying data files (or file parts) at the operating system level?
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
Hi Marcus,
To clarify, are you copying data files (or file parts) at the operating system level?
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
To clarify, are you copying data files (or file parts) at the operating system level?
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
That's the the plan, yes.
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
FYI, It is possible that if the production data files are being written to while you are copying at the OS level, the copy could get corrupted. This could be avoided by pausing the database at off-hours or recently in the case of a system that is busy 24/7, you could create Q-pointers to the files and write a program or paragraph to select/copy the records file-by-file within Universe.
Not knowing your environment or if you considered this already, just tossing in my 2 cents :-)
Regards
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
Marcus,
FYI, It is possible that if the production data files are being written to while you are copying at the OS level, the copy could get corrupted. This could be avoided by pausing the database at off-hours or recently in the case of a system that is busy 24/7, you could create Q-pointers to the files and write a program or paragraph to select/copy the records file-by-file within Universe.
Not knowing your environment or if you considered this already, just tossing in my 2 cents :-)
Regards
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
FYI, It is possible that if the production data files are being written to while you are copying at the OS level, the copy could get corrupted. This could be avoided by pausing the database at off-hours or recently in the case of a system that is busy 24/7, you could create Q-pointers to the files and write a program or paragraph to select/copy the records file-by-file within Universe.
Not knowing your environment or if you considered this already, just tossing in my 2 cents :-)
Regards
------------------------------
Mark Sapp
Senior Software Developer
Rocket Software
Lee's Summit, Missouri USA
------------------------------
------------------------------
Peter Cheney
Rocket Forum Shared Account
------------------------------
Provided you have sufficient disk space the if OS (HPUX?) supports snapshot of the file system(s), would it be feasible to pause the database, snap the filesystem(s), resume the database and then copy the files at OS level? Scripting the pause/snap/resume would likely be over in an acceptably short time.
------------------------------
Peter Cheney
Rocket Forum Shared Account
------------------------------
------------------------------
Peter Cheney
Rocket Forum Shared Account
------------------------------
- The donor location must be quiescent - i.e. database paused or stopped
- The recipient location must either be new and empty, or it should be verified that the UniVerse instance on the server concerned is not using ANY of the files that are to be replaced. (see 'handle' on Windows or 'fuser -u' on UNIX).
CAUTION: If using Automatic Data Encryption (ADE), please refer to the ADE documentation as specific steps are advised when copying ADE encrypted data to ensure the metadata files are not compromised.
------------------------------
John Jenkins
Principal Technical Support Engineer
Rocket Software Limited
U.K.
------------------------------
We have test and production accounts on the same system, which is to say that they reside in different paths on HP-UX.
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
When the test account was first created, it was a clone of the production system with adjustments in various places to reflect the different paths.
But our main files, ORDERS, PAYMENTS, and some others, are distributed files, and the distribution algorithms are equivalent on both test and production, as are the number of partitions.
So, here are the questions:
a) Can I refresh the data on test by simply copying the individual data portions from production to test?
b) I recently fixed one broken file by copying the affected partfile from production to test, but maybe I didn't really. It seems to work, but maybe I'm missing something. I don't know for sure. How can I be sure?
c) And what about the stub files? Are there even stub files on UV? Do they contain the paths in which they were configured?
d) Where can I find the latest, on-line documentation about this?
------------------------------
Marcus Rhodes
Owner
Rocket Forum Shared Account
------------------------------
The path to [file]'s coresponding I_[file] index dir is buried inside the header of the file being indexed. So if you use OS level tools to copy from prod to dev, the prod indexes may get updated from dev.
SET.INDEX's INFORM option will tell you.
SET.INDEX's RELATIVE.PATH option will help you avoid the problem altogether.
------------------------------
Chuck Stevenson
DBA / SW Developer
Pomeroy
San Ignacio US
------------------------------
Sign up
Already have an account? Login
Welcome to the Rocket Forum!
Please log in or register:
Employee Login | Registration Member Login | RegistrationEnter your E-mail address. We'll send you an e-mail with instructions to reset your password.
77 4th Avenue
Waltham, MA 02451 USA
