Question Steps to Move OpenEdge 10.2A Database Files to New Storage

Mike

Moderator
Hi Team,
We are planning to move our OpenEdge 10.2A database files to new storage on the same server.
The database extents are currently located across /datos, /indices, /bimage, and /aimage.
Could you please provide the step-by-step procedure for this activity, including:

- Database and application shutdown steps
- Safe method to move the data, index, BI and AI extents
- Changes required if the mount-point names change
- Database and application startup steps
- Post-migration validation
- Rollback steps
- Expected downtime
Please also let us know what database and server details you need from us to prepare the procedure.

Multi-user session begin.
[2019/06/15@05:45:04.980-0400] P-25810 T-1 I BROKER 0: (4234) Progress OpenEdge Release 10.2A build 1185 on HP-UX sapian B.11.31 U.
[2019/06/15@05:45:04.980-0400] P-25810 T-1 I BROKER 0: (4281) Server started by root on batch.


# proutil -C dbipcs
OpenEdge Release 10.2A as of Fri Oct 2 21:05:05 EDT 2008

PROGRESS SHARED MEMORY STATUS
ID ShMemVer Seg# InUse Database
0 - - - (not PROGRESS)
1 - - - (not PROGRESS)
2 - - - (not PROGRESS)
32771 - - - (not PROGRESS)
4 - - - (not PROGRESS)
5 - - - (not PROGRESS)
6 6412003 0 Yes /data/ladb3/ladb3.db
7 6412003 0 Yes /data/ladb4/ladb4.db
8 6412003 0 Yes /data/lagdb2/lagdb2.db
9 6412003 0 Yes /data/help1/help1.db
10 - - - (not PROGRESS)
11 - - - (not PROGRESS)


/data/ladb3/ladb3.db :- 35 GB size
/data/ladb4/ladb4.db 20gb
/data/lagdb2/lagd2.db 88157 MB
/data/help1/help1.db 267 mb

Enter your selection: q
# df
/aimage (/dev/vg01/lvol4 ): 56590590 blocks 7545411 i-nodes
/apps (/dev/vg02/lvol4 ): 9452446 blocks 1260155 i-nodes
/backups (/dev/vg01/lvol3 ): 87455358 blocks 11660679 i-nodes
/bimage (/dev/vg02/lvol5 ): 56029142 blocks 7470549 i-nodes
/ctry (/dev/vg02/lvol3 ): 6325180 blocks 817410 i-nodes
/datos (/dev/vg00/lvol9 ): 85939802 blocks 11458608 i-nodes
/home (/dev/vg00/lvol5 ): 154336 blocks 8117 i-nodes
/indices (/dev/vg01/lvol2 ): 91926878 blocks 12256914 i-nodes
/olap (/dev/vg02/lvol2 ): 28484742 blocks 3797259 i-nodes
/opt (/dev/vg00/lvol6 ): 3434832 blocks 54069 i-nodes
/pruebas (/dev/vg02/lvol1 ): 97539400 blocks 13005212 i-nodes
/tmp (/dev/vg00/lvol4 ): 448304 blocks 7038 i-nodes
/usr (/dev/vg00/lvol7 ): 7437536 blocks 117115 i-nodes
/var (/dev/vg00/lvol8 ): 4768848 blocks 74864 i-nodes
/stand (/dev/vg00/lvol1 ): 2877248 blocks 45300 i-nodes
/ (/dev/vg00/lvol3 ): 2005936 blocks 31587 i-node



# proutil /data/lagdb3/lagdb3 -C describe
OpenEdge Release 10.2A as of Fri Oct 31 21:05:05 EDT 2008


OpenEdge Database Description

Database Name : /data/lagdb3/lagdb3
Version : 150.0
Block Size : 8192
Largest Cluster : 64
Create Date : Fri Jun 19 23:36:13 2015
Last Open Date : Sat Nov 15 09:31:02 2025
Prior Open Date : Sat Nov 15 09:31:02 2025
Schema Change Date : Thu Jan 4 15:20:39 2018

Before Imaging information
Block Size : 8192
Cluster Size (16K Units) : 32
Last Open Date : Sat Nov 15 09:31:03 2025

Backup Information
Last Full Backup Date : Sat Nov 15 03:55:00 2025
Last Incremental Backup : *** Not yet performed ***

Database Features

ID Feature Active Details
---- --------------------------------- ------ -------
5 Large Files Yes
9 64 Bit DBKEYS Yes
10 Large Keys Yes
11 64 Bit Sequences Yes



# cat help10.st
#
b /bimage/lahelp/ f 100096
b /bimage/lahelp/
#
d "Schema Area":6,64;1 .
#
d "Common_Area":7,64;64 . f 200064
d "Common_Area":7,64;64 .
#
d "indices":8,64;64 /indices/lahelp/ f 100096
d "indices":8,64;64 /indices/lahelp/
#
a /aimage/lahelp/ f 100096
#
a /aimage/lahelp/ f 100096
#
a /aimage/lahelp/


# cat ladb1.st
#
b /bimage/ladb1 f 100000
b /bimage/ladb1 f 100000
b /bimage/ladb1
#
d "Schema Area":6,64;1 .
#
d "Lower_Area":7,128;64 . f 1536000
d "Lower_Area":7,128;64 . f 1536000
d "Lower_Area":7,128;64 .
#
d "Low_Area":8,64;64 . f 3048000
d "Low_Area":8,64;64 .
#
d "Medium_fcs_sum_Area":9,64;64 . f 2000000
d "Medium_fcs_sum_Area":9,64;64 .
#
d "Medium_idh_hist_Area":10,32;64 . f 2560000
d "Medium_idh_hist_Area":10,32;64 .
#
d "Medium_trgl_det_Area":11,128;64 . f 5000000
d "Medium_trgl_det_Area":11,128;64 .
#
d "Heighter_tr_hist_Area":12,64;512 . f 5000000
d "Heighter_tr_hist_Area":12,64;512 .
#
d "Common_Area":13,128;64 . f 3536000
d "Common_Area":13,128;64 .
#
d "Index_area":14,1;8 /indices/ladb1 f 4000000
d "Index_area":14,1;8 /indices/ladb1
#
a /aimage/ladb1 f 200064
a /aimage/ladb1 f 200064
a /aimage/ladb1 f 200064
a /aimage/ladb1


# uname -a
HP-UX sapian B.11.31 U ia64 4209233176 unlimited-user license
#
 
1) Create a .st file with your desired new storage in the new db home directory. Use absolute path names, do not use relative names like ".".

2) Make a backup

3) Restore using the new .st file

Feel free to practice using copies of the sports database. It is a quick and effective way to verify that you are doing things correctly.
 
1) Create a .st file with your desired new storage in the new db home directory. Use absolute path names, do not use relative names like ".".

2) Make a backup

3) Restore using the new .st file

Feel free to practice using copies of the sports database. It is a quick and effective way to verify that you are doing things correctly.
I would like to clarify our complete migration objective and the testing already completed.
The target storage is not another local filesystem or a SAN/block device. It is an NFSv3 export from a separate NAS server. The NFS export is mounted on the existing HP-UX 11.31 server where the OpenEdge 10.2A database brokers and QAD processes run.
Our current production environment uses local HP-UX logical volumes for the following filesystems:
  • /apps for the OpenEdge installation and application components
  • /matos for database control and data files
  • /Slices for index extents
  • /bim for BI extents
  • /aim for AI extents
  • /backups for database backups
  • /ctry, /olap, and /pruebas for QAD, reporting, and related application files
Our actual production plan is not currently to restore every database into a newly named structure.
The intended production approach is:
  1. Create matching directories on NFS for the existing application and database filesystems.
  2. Perform an initial filesystem-level copy from the local logical volumes to NFS while preserving ownership, group, permissions, setuid/setgid bits, timestamps, symbolic links, hard links, executable files, and large files.
  3. During the approved outage, stop QAD users, interfaces, batch processes, reporting and OLAP jobs, and all OpenEdge databases.
  4. Perform a final synchronized copy while all application and database activity is stopped.
  5. Verify file counts, sizes, ownership, permissions, links, and critical checksums.
  6. Unmount the existing local filesystems without immediately deleting or reformatting the local logical volumes.
  7. Mount the NFS directories using the same original mount-point names, including /apps, /matos, /slices, /bim, /aim, /backups, /ntry, /slap, and /prbaas.
  8. Start the OpenEdge databases and QAD application from their existing paths.
  9. Complete database, application, batch, interface, backup, recovery, performance, and business validation.
  10. Retain the original local logical volumes during an agreed rollback and stabilization period.
  11. Permanently remove the local logical volumes only after successful validation and formal approval from the DBA, UNIX/Storage, QAD Application, and business teams.
Because the final mount-point names will remain unchanged, the OpenEdge DLC path will remain:
/apps/pro102a/dlc
The database extent paths will also remain unchanged:
  • /matos
  • /slices
  • /bim
  • /aim
Our understanding is that PROSTRCT REPAIR should not be required for the final production cutover if every pathname remains exactly the same.
We have already completed two isolated database tests on NFS.
Test 1: Physical extent copy and PROSTRCT REPAIR
  • We created a restored test database on the existing local filesystems.
  • We stopped the test database cleanly.
  • We physically copied its control, data, index, BI, and AI extents to an isolated NFS directory.
  • We compared the source and NFS files using checksums and byte counts. All copied files matched.
  • We created an updated .st file containing fully qualified absolute NFS paths.
  • We ran PROSTRCT REPAIR against the copied database control file.
  • We verified with PROSTRCT LIST that all database extents pointed to the NFS paths.
  • The copied database started successfully in multi-user mode from NFS.
  • Shared memory was created and database connections were enabled.
Test 2: Backup and restore directly to NFS
  • We took an OpenEdge backup and restored it directly into an isolated NFS directory.
  • The restored database was created successfully on NFS.
  • The restored database started successfully in multi-user mode.
  • This confirmed that the backup, .st, and restore method also works technically in our test environment.
We also completed the following NFS backup tests:
ladb1 test
  • Online backup completed successfully.
  • Final backup size was approximately 21.87 GiB.
  • Backup duration was 32 minutes and 10 seconds.
  • PROREST -vp verification completed successfully.
  • 2,875,164 database blocks were verified.
  • Verification duration was approximately 3 hours.
lagdb test
  • Online backup completed successfully.
  • Backup size was approximately 3 GiB.
  • Backup duration was 4 minutes and 7 seconds.
  • PROREST -vp verification completed successfully.
  • 393,155 database blocks were verified.
  • Verification duration was 5 minutes and 42 seconds.
During database startup from NFS, OpenEdge logged message 9466 for the database files because they were located on a remote device.
However, after message 9466:
  • Multi-user startup completed successfully.
  • Shared memory was created successfully.
  • Database connections were enabled successfully.
  • No missing-extent, permission, locking, or database I/O errors were observed during the startup tests.
We understand that message 9466 is informational and will continue to appear whenever active database files reside on NFS, regardless of whether the database is physically copied, repaired with PROSTRCT REPAIR, restored from backup, or created directly on NFS.
We also observed that some NFS filesystem and metadata operations responded slowly during sustained backup and verification activity. For example, detailed file-listing commands occasionally waited while a large backup was being written. The backup still completed successfully, but this latency will require review by the UNIX/Storage team before production approval.
The OpenEdge installation and QAD application components under /apps may also be moved to NFS. The OpenEdge and QAD processes will continue to execute on the HP-UX server, but their executable files will be read through the NFS-mounted /apps filesystem.
We therefore also need to validate:
  • OpenEdge executables from the NFS-mounted /apps
  • _dbutil ownership and setuid permissions
  • NFS noexec and nosuid behavior
  • Symbolic and hard links
  • Shared-library loading
  • QAD application startup and shutdown
  • OLAP and batch processing
  • Interface processing
  • Application transaction response time
  • Database backup and recovery performance
  • NFS locking, caching, synchronous-write behavior, and failure handling
For production, we are currently considering the filesystem-copy and same-mount-point approach rather than restoring every database individually.
Could you please advise on the following?
  1. Is the filesystem-level copy and same-mount-point cutover approach safe for OpenEdge, provided that all databases, QAD processes, interfaces, batch jobs, and other writers are fully stopped during the final synchronization?
  2. If all absolute database extent paths remain unchanged after the NFS mounts replace the local filesystems, do you agree that PROSTRCT REPAIR is not required?
  3. Would you recommend your verified backup, absolute-path .st, and restore method instead of the complete filesystem-copy approach for the production databases?
  4. Do you consider permanently running active OpenEdge 10.2A database, index, BI, and AI files on NFSv3 under HP-UX 11.31 to be a supported and safe production design?
  5. Are there specific requirements regarding NFS mount options, file locking, synchronous writes, write ordering, attribute caching, buffered I/O, hard mounts, transport protocol, or NAS/network failure handling?
  6. Would you recommend keeping /apps/pro102a/dlc on local storage rather than moving the OpenEdge binaries and QAD application files to NFS?
  7. Are there any additional tests that you would recommend before the business permanently uses the NFS-based databases and the old local logical volumes are removed?
The old local logical volumes will not be removed immediately after cutover. They will remain intact and available for rollback until database validation, QAD functional testing, OLAP and batch validation, backup and recovery testing, performance review, and formal approvals are completed.
Thank you for your guidance.
 
So your actual project is quite a bit different from your original question.

As you very well know by now 10.2A is unsupported. Given that you're sticking with OpenEdge 10.2A is it also safe to assume that you are using equally ancient and unsupported versions of HPUX and NFS? Is the hardware as old and unsupported as everything else?

Prudence would dictate that you ought to be figuring out how to migrate this environment to supportable infrastructure.

Putting databases on NFS volumes is a really bad idea. That message isn't just for decoration. Sure, it will "work" right up until the point where it stops working. You're really setting yourself up for some serious performance performance and reliability issues. Maybe that's ok if this is a reference copy of a system that is only accessed by a few people once in a blue moon to check old data or something like that. But if it is a real production system with users that expect performance and reliability you won't be delivering a very good solution.
 
Thanks Tom. I understand your concerns, especially around running an old OpenEdge 10.2A database on NFS.
Just to provide some additional context, this environment has very few active users and the workload is quite low. At the moment, we are only testing a copy of the database on NFS to understand the feasibility and behavior before making any decision for the actual environment.
We will also consider the performance and reliability risks you mentioned. One point I would like to clarify: when you recommend migrating to supported infrastructure, do you mean moving the entire environment to a newer supported OpenEdge/OS/storage platform rather than moving the existing 10.2A database to NFS?
Also, considering that this is a very low-usage environment, would you still strongly advise against NFS, or could it be considered with proper testing? If there are any specific OpenEdge/NFS risks or tests you would recommend, that would be very helpful.
Thanks again for your guidance
 
Back
Top