Workspace continually refreshed, Savepoints & MergeWorkspace scenario

Hi,
I'm using Oracle WorkspaceManager from oracle version: Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bit
1./ I create a workspace with the isRefreshed flag set to true.
2./ I immediately create a savepoint A.
3./ I insert a row in a versioned table.
4./ Create savepoint B.
4./ exec dbms_wm.setdiffversions('workspace', 'A', 'workspace', 'B')
5./ Querying the diff view should show me 3 rows.
DiffBase and workspace, A rows would have wmDiffCode: NE (Not existing)
workspace, B row would have wmDiffCode: I (insert)
All is fine so far.
5./ I merge the workspace to LIVE
6./ I query the diff view again:
DiffBase and workspace, A rows would have wmDiffCode: NC (Not changed)
workspace, B row would have wmDiffCode: I (insert)
I confirm that step 6./ shows wmDiffCode: NE if I disable continual refresh.
Can somebody tell me why the wmDiffCode changes to NC in this scenario? I thought that creating a savepoint was saving a snapshot of the workspace (including DIFF view states) that should not be alterable by any refresh operations on the workspace, it seems like this assertion is false.
Any insights appreciated,
Thanks,
-Guillaume Jeudy

Hi,
This is expected. The NE code indicates that the row does not currently exist as of that version/savepoint. However, when the row is merged into the LIVE workspace, and due to the fact that the child workspace is being continually refreshed, this new row will become visable from savepoint A. Any changes into the LIVE will automatically refresh any child workspaces(including earlier savepoints) up to the point where each row is further modified. Since the row is now visible from savepoint A, the code needs to changed from NE to NC.
When the workspace is not being continually refreshed, the merge into the LIVE workspace does not cause any changes to savepoint A, and so the row remains not existent.
Regards,
Ben

Similar Messages

Is it possible to "freeze" a continually refreshed workspace?

Hello,
I'm developing a project management system based on workspaces.
The aim is to create and develop projects inside the workspaces, parallel to the stream LIVE, and after approvall, do the merge.
I have two requirements:
Req1 - Workspaces should be continually refreshed, during their "livetime".
Req2 - After the merge, it is necessary to save the final information in the workspace, just for future read.
I thought the following approach:
Use workspaces in CR mode and after merge, switch to NCR. The scenario would be stable and unchanging, maintaining the necessary historical data.
I found out that it's not possible to change a CR workspace to a NCR.
Besides, the raising of workspaces (tried with 30000), cause general database degradation. Queries on version-enabled and merge/refresh operations were getting slower and slower.
Three questions about this:
1 - Is there any way to "freeze" CR workspaces? Like... avoid the auto-refresh process?
2 - Is it natural that queries and merge / refresh operations become successively slower with increasing workspaces? (i noticed that many wmsys.wm$ tables/views are used in these processes and it seems to me that their performance is getting worse. Is that possible? If so, is there any way to optimize?)
3 - Assuming that the use of thousands of workspaces may have some performance issues, what is the best way to save a final (total and unchangeable) "image" of the database after doing the merge of a project?
NOTE:
I thought using a system with just one CR workspace (history WS) and savepoints... just to keep the final views of the projects.
After merging to LIVE, the history workspace would be auto-refreshed, I would put a savepoint there and keep the final vision. But with a CR workspace, the savepoint doesn't work in the way i need... the vision is altered every time we change rows in LIVE, right?
So, i tried to use a NCR workspace. After merge operation, i force a refresh in the history workspace, put the savepoint and it's done. But there's still a little problem with this scenario: LIVE may have hundreds or thousands of changes to refresh... the process can be slow and block the LIVE records for a long time.
Am i jumping to conclusions or assuming something wrong? Is there any technical solution that I'm not seeing?
Any help or ideas are welcome!
Thanks in advance.
Best regards,
Pedro Lourenço

Hi,
To answer your questions:
1. There is a way to freeze a workspace, using dbms_wm.FreezeWorkspace, but it would only prevent dmls / workspace operations from taking place within the child workspace. It would not remove the auto-refresh functionality of the workspace.
2. No, you should not being having a significant performance impact on queries by just adding workspace. Have the statistics of the WMSYS schema been gathered, and for the schemas that contain the version enabled tables? If not, I would recommend doing so. If you have done so already, I would need to see specific queries, etc.
3. If you need to get a snapshot of the data without the possibility of it being changed, you can't use a CR workspace, so your only option would be a NCR one. How much of the data that is being modified by a child workspace is being additionally modified by other children workspaces? Essentially, does each workspace work with its own distinct data or would there be alot of overlap(ie conflicts) between them? How many of these workspaces need to be merged and/or refreshed simultaneously?
Have you thought about creating a savepoint in the LIVE workspace at the time of the merge and use that as the snapshot? It would not work in all use cases, so would depend on the specifics of your setup. I'll continue thinking if there are existing ways to accomplish this.
Also, in regards to CR workspace savepoints, the savepoint data can be changed if the child workspace doesn't have its own version of the row. Any new row in LIVE would override the view of the data from each savepoint.
Regards,
Ben

Update in continually refreshed workspace

Hi,
From experiment I can see that when row is updated in child continually refreshed workspace it stops receiving updates made in parent workspace (which makes sense). What should happen to for that row to start receiving updates again? I tried refresh and merge and it doesn't seem to work.
Thanks,
Artem.

Hi Artem,
As long as a row exists in the child workspace, no automatic updates will be done to that particular row. So, the only way to reinstate this is to remove it. So, without recreating the workspace, this can be done by using MergeTable with the remove_data parameter set to true, or by executing RollbackWorkspace, RollbackTable, or RollbackToSP to remove all of the changes done to the row in the child workspace.
Regards,
Ben

FF 4Beta seems to be continuously refreshing itself 2-5 times per second - how to stop this ?

FF 4Beta seems to be continuously refreshing itself 2-5 times per second - how do I stop this from occurring.

FF 4Beta seems to be continuously refreshing itself 2-5 times per second - how do I stop this from occurring.

Mail and Yahoo! Mail continually "refreshing"

I recently added my Yahoo! mail account to Mac OS X mail. The problem is that it is continually "refreshing". The wheen for the Yahoo! account never stops. None of the other mail accounts I have set up are having this issue so I'm pretty sure it is a problem with the Yahoo! account but I don't know if it is something in the settings on the Yahoo! end that I can and need to change, or if there is a setting on the Mac OS X Mail setting I can and should change. Has anyone else had this issue? If so, have you solved it, and if so, how? Using 10.7/Lion.

Gmail, AOL, almost everything but Yahoo!. It ***** because I use my Yahoo! account for Facebook so that I get them in realtime on my phone but I can use it in Mail. This really aggravates me...a lot.

Safari webpage continuous refresh

I'm trying view www.realtor.ca, which provides real estate listings for Canada. After putting my search criteria it locates the listings but then continuously refreshes the page before I even get a chance to review the results. What's up with this annoying bug?

The site refreshes on two different browsers on my iPad - not just Safari. I'd guess that it's not an iPad friendly site and not a Safari issue.

In HOTMAIL Fireefox Beta continually refreshes automatically. How to stop the refresh?

In hotmail.com the screen automatically continues to refreshes resulting in the displayed screen continuously reloading every second. How to stop this automatic refresh?

See [/questions/778953]

LR 4.3 continuously refreshes the image in develop mode

I am using LR 4.3 with Photoshop CS5. I have found that after editing a picture in photoshop (transfer from CR2 to .TIFF file) it correctly transfers back into LR as an updated .TIFF file with all edits from PS. However, on further editing of the picture in LR the image is refreshed after EACH edit (ie. file reloads into LR taking a few seconds). If I am trying to apply an edit as the refresh starts then that edit gets lost i.e. NOT applied. The refresh happends after ANY change to the file. It is very frustrating as it wastes time and makes editing slow (have to keep waiting for a refresh). It does not appear to happen in that way when editing CR2 files.
I have only noticed this feature since installing V4.3 although it may have been happening in previous versions. Is there any way round this problem?
I am using W7, i5 2.5 Ghz processor and 8 GB ram with dual screens.

This problem has existed in LR since about 3.0.
If Lightroom creates a tiff or psd for external edit, then when it is returned to LR and one attempts to further adjust the image, the adjustment sliders gray out, the photo reloads, then the sliders become active again.
Many folks say it is caused by the program writing to XMP, to turn "automatically write to XMP" off. Does NOT work! It is not the root cause.
Bampton says to flick to the next photo. Does NOT work. This is not normal behavior.
What will work to stop the activity is to close and reopen the program. This is not normal behavior.
This will not occur if the file(tiff or PSD) is written by Photoshop. If you are using 4.X and CS6 then Photoshop will write the file at the end of your session in PS. No blinkies.
If you are using 3.X and CS5 together you will not experience the problem since again PS writes the file. No blinkies.
If you are using 4.x and CS5 or under you will encounter the problem, blinkies!
I have not encountered the problem with any file written by the OS or any other program.
The problem is ONLY with tiffs or PSD files created by LR for external editing. It would seem to me this is a bug.
It is not a major bug, but annoying.
I have filed a bug report again here......http://feedback.photoshop.com/photoshop_family/topics/lr_continuously_refreshes_the_image_ in_develop_mode?rfm=1
Please drop by and sign it...........maybe we can get some attention this time. I filed a bug report before, but can't find it.
Thank you for your time,
Rich

Continuously refreshing a tab after an interval leads to high memory consumption (400MB to 800MB in 30 seconds for 3 refreshes at 10 secs interval), why?

Environment:
MAC OSX 10.9.5
Firefox 32.0.3
Firefox keeps consuming lot of memory when you keep refreshing a tab after an interval.
I opened a single tab in my firefox and logged into my gmail account on that. At this stage the memory consumption was about 400MB. I refreshed the page after 10 seconds and it went to 580MB. Again i refreshed after 10 seconds and this time it was 690MB. Finally, when i refreshed 3rd time after 10 seconds, it was showing as 800MB.
Nothing was changed on the page (no new email, chat conversation, etc. nothing). Some how i feel that the firefox is not doing a good job at garbage collection. I tested this use case with lot of other applications and websites and got the similar result. Other browsers like Google chrome, safari, etc. they just work fine.
For one on of my application with three tabs open, the firefox literally crashed after the high memory consumption (around 2GB).
Can someone tell me if this is a known issue in firefox? and is firefox planning to fix it? Right now, is there any workaround or fix for this?

Hi FredMcD,
Thanks for the reply. Unfortunately, i don't see any crash reports in about:crashes. I am trying to reproduce the issue which will make browser to crash but somehow its not happening anymore but the browser gets stuck at a point. Here is what i am doing:
- 3 tabs are open with same page of my application. The page has several panels which has charts and the javascript libraries used for this page are backbone.js, underscore.js, require.js, highcharts.js
- The page automatically reloads after every 30 seconds
- After the first loading of there three tabs, the memory consumption is 600MB. But after 5 minutes, the memory consumption goes to 1.6GB and stays at this rate.
- After sometime, the page wont load completely for any of the tabs. At this stage the browser becomes very slow and i have to either hard refresh the tabs or restart the browser.

Menu bar icons continually refreshing

Every 15 seconds or so, all of the items on the right side of menu bar - including the clock, but excluding the Notification Center icon - disappear and rebuild themselves. This started yesterday when I upgraded to OS X 10.8.2. I've removed all of the third party apps from the menu bar and it is still happening.
Any help would be greatly appreciated.
Thanks,
Troy

disable the spotlght icon. the service will still run in the background. you're just hidng the icon.
Hide the Spotlight Menu Icon in OS X
To reemphasize, this does not disable Spotlight or mds, it only hides the icon from the menubar.
Launch the Terminal from /Applications/Utilities/ and type the following command string exactly:
sudo chmod 600 /System/Library/CoreServices/Search.bundle/Contents/MacOS/Search
Hit return, and then you will need to kill a process called “SystemUIServer” to refresh the menubar in OS X and have the change take effect:
killall SystemUIServer

Continuously Refreshing Apple screen

I just updated my 5G 60 gig iPod to the new firmware. I was able to use my iPod perfectly up until the disconnection from the computer. My iPod displayed the "Plug into the wall" screen. That's a little odd seeing as how my iPod is a 5g. It usually only does that for 4G and below. But now that I plugged it into the wall, all it does is go to the black screen with a silver apple on it, reset itself and go right back to it.
Could my iPod have been updated with the fourth generation's firmware on accident? I really want it fixed again!

****!
The iPod finally died from nonstop hard disk use. I plugged it into the computer and an old iPod photo non-color charging icon. You know, the old
icon except without going past the first part.
My iPod must have been flashed with the wrong firmware. Any idea on how to get it back to iPod 5G firmware?

Automatically refreshed workspace and changes into LIVE

Hi,
I've noticed that changes made directly into LIVE
are immediately visible to automatically refreshed
child workspaces. However the answer to the FAQ
question "When sharing changes made in a child
workspace with another child workspace, should I
use periodic (manual) or continually (automatic)
refresh?", states that this type of change must
be refreshed manually.
What is the expected behaviour afterall?
Regards,
Karina Borges

Karina,
For continually refreshed workspaces, changes made in the LIVE workspace will be immediately visible.
For non-continually refreshed (NCR) workspaces, the changes are not immediately visible - to see the changes, NCR workspaces have to be manually refreshed.
We will update the FAQ to make this point clearer.
regards,
- Ramesh.

Workspace refresh cascading

Hi there,
We're trying to create a setup where we have a linear hierarchy of workspaces (A is a child of LIVE, B is a child of A) that would allow us to phase propagating the changes from the LIVE environment to the other workspaces in a controlled fashion.
However, we're noticing something that does not work as expected (based on our understanding of the documentation): when refreshing A from LIVE, the changes make it into B, too. Since we're refreshing only A (and none of the workspaces are created with the continuously refresh mode), we'd expect the changes to stop at A and to only move to B when B is explicitly refreshed.
Are we missing something fundamental here? We checked the workspaces creation documentation several times, but it does not mention the changes being propagated to the ancestors when refreshing a workspace, so this looks like a bug to us. We're using 10.2.0.4.3, BTW.
Please advise.
Best regards,
Babacu

Hi,
This is expected behavior. When using the default behavior of dbms_wm.RefreshWorkspace, the parent version of the workspace being refreshed is modified to point to the current version of its parent workspace. This will cause any child workspaces to be refreshed at the same time. If you only would like to refresh the one workspace, and not any of its children, then you can set the copy_data parameter to true, which will copy all of the required rows into the workspace being refreshed, and leave its children workspaces unaffected.
Regards,
Ben

Error while refreshing bqy file on workspace using stored procedure

Hi,
I am using stored procedure which returns ref-cursor. The report is running fine locally but on Workspace, while refresh, it gives the following error.
An Interactive Reporting Service error has occurred.-SQL API: [SQLNumResultCols], SQL RETURN: [-1], SQL STATE: [HY010], SQL NATIVE ERROR: [0], SQL MESSAGE: [[Microsoft][ODBC Driver Manager] Function sequence error]
(0)
I have ODBC/ODBC connection set up at BI+ configurator.
Thanks,
Manish

We also met this issue. Some guy told me this is a Hyperion limitation, is it true? I really need this function in Hyperion:(

Long lasting Workspaces with many Rows, Merging and Locks

Hi all!
We sucessfully implemented a solution using Oracle Workspace Manager combined with Oracle Locator (Spatial). The present version of our product uses Workspace Manager to enbale "private working areas" for the users. They can edit data in this areas (workspaces), goto savepoints (undo) and merge their work with the live-data. The live-data meanwhile is locked for other users.
Our current workspaces have a short lifetime and contain about 100-200 versioned datarows. A merge operation takes about 5 to 10 seconds.
We are planing to enhance our system in terms of long time planing situations. The user will be able to edit a project (workspace) and keep his editings for a long time. So what we will have are:
- Workspaces that exist for a long time and
- Workspaces that include thousands of versioned rows.
Any experience with this kind of Workspace Manager usage? What about the merging performance? Do merging or refesh operations block each other? Which locks are set? Unfortunately I can't find any hints in the documentation.
Please help, any opinions welcome!
Thanx,
Nothi

Hi,
This is a common usage for Workspace Manager, so since you already have a working solution you should not have any issues when using workspaces that exist for a long period of time. In terms of the actual performance, it would be dependent on the number of rows being merged, the total size of the base table among other things, etc and so I will not be able to exactly qualify the difference between your two cases.
You can merge and refresh unrelated workspace simultaneously. However, if you are merging and refreshing the same workspace or its parent at the same time, locks will needed to be taken to guarantee the validity of the data. If you are running into locking issues, you may be able to use the row_level_locking system parameter, depending on your exact situation.
Without changing the value of this system parameter(default is false/off), a MergeWorkspace will acquire exclusive locks on the parent and child workspaces, while a RefreshWorkspace will also acquire an exclusive lock on the parent and a shared lock on the child. After setting row_level_locking to true, the child workspace locks will remain the same, but the parent workspace lock for either MergeWorkspace or RefreshWorkspace will become a shared lock along with the necessary row locks needed by the Merge operation.
Hope this helps.
Ben
null

Workspace continually refreshed, Savepoints & MergeWorkspace scenario

Similar Messages

Maybe you are looking for