Page 2 of 4

Re: Cryptic TM1 Messages Decoded

Posted: Tue Jul 12, 2011 1:16 pm
by John Hammond
TM1 Invalid key: Dimension Name: "Source", Element Name (Key):

You EXECUTEPROCESS'ed a job in the Prolog that relied on Elements added in the Prolog. Remember any dimension changes are not committed until you exit the prolog, so putting your calls in the epilog will resolve the issue.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Aug 03, 2011 8:36 am
by John Hammond
RESTRICTION: - Function "ExecuteProcess" not valid for Tm1 datasource

You can't ExecuteProcess from Metadata or Data tabs.

Re: Cryptic TM1 Messages Decoded

Posted: Thu Aug 18, 2011 10:57 am
by John Hammond
tm1 Cannot get Process Handle

You lost connection to server and need to refresh and login again.

Re: Cryptic TM1 Messages Decoded

Posted: Fri Aug 19, 2011 5:09 am
by Steve Rowe
Hi John,
You can use ExecuteProcess in the data and metadata tabs, I've done it alot on many occasions. There must be another or different problem.
Cheers

Re: Cryptic TM1 Messages Decoded

Posted: Thu Aug 25, 2011 7:46 am
by John Hammond
cheers Steve - will take a look as to exactly why this occurred though in this case removing the executeprocess cured the problem.

Message: TM1 Rule Creation Failed

Reason: You unloaded a cube when you were in the rules editor and it lost its connection to the cube.

Re: Cryptic TM1 Messages Decoded

Posted: Fri Aug 26, 2011 12:37 am
by Andy Key
John Hammond wrote:RESTRICTION: - Function "ExecuteProcess" not valid for Tm1 datasource

You can't ExecuteProcess from Metadata or Data tabs.
This is a version related issue, and the error message is correct. It used to be the case that you couldn't use ExecuteProcess on the Metadata or Data tabs with a CubeView or DimensionSubset data source. Whether it was 9.1 or 9.4 that first corrected this I don't recall, but as Steve says, it's not an issue any more. Well, unless you are still on version 9.0 of course...

Re: Cryptic TM1 Messages Decoded

Posted: Fri Aug 26, 2011 12:53 am
by Alan Kirk
Andy Key wrote:
John Hammond wrote:RESTRICTION: - Function "ExecuteProcess" not valid for Tm1 datasource

You can't ExecuteProcess from Metadata or Data tabs.
This is a version related issue, and the error message is correct. It used to be the case that you couldn't use ExecuteProcess on the Metadata or Data tabs with a CubeView or DimensionSubset data source. Whether it was 9.1 or 9.4 that first corrected this I don't recall, but as Steve says, it's not an issue any more. Well, unless you are still on version 9.0 of course...
Actually... it was changed functionality in 9.0 SP3 and as you indicated only ever applied to cube view or subsets as the data sources. It works just fine and dandy in 9.0 SP3 U9 with an ASCII data source.

From the 9.0 SP3 U1 Release Notes:
When a TurboIntegrator process uses either a TM1 cube view or TM1 subset as a datasource, the ExecuteProcess function is valid only in the Prolog or Epilog procedures.

This is a departure from previous versions of TM1, where ExecuteProcess could be called from any procedure within a TurboIntegrator process when the datasource was a TM1 cube view or TM1 subset. Applix introduced this change in functionality to ensure data integrity and prevent server crashes that can result from locking conflicts that arise when a process calls the ExecuteProcess function from the Data or Metadata procedures.

If you must use the ExecuteProcess function from the Data or Metadata procedures when a process datasource is a TM1 cube view or TM1 subset, Applix suggests the following solution:
1. Create an initial TurboIntegrator process to export the TM1 cube view or subset to an ASCII file, using the ASCIIOutput function.
2. Create a second TurboIntegrator process using the resulting ASCII file as a datasource. You can then call the ExecuteProcess function from any procedure within the process when the datasource is an ASCII file.
3. Combine both processes in a chore.
Given the change in the locking model in 9.1 it's possible that they revoked this change, though personally I've never had cause to use it in those circumstances.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 10:06 am
by John Hammond
Thanks Alan - your diagnosis on the versions is correct.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 10:08 am
by John Hammond
TM1 cannot find change I know I made in the log

Your start date was prior to the earliest existing log file. Change it to match the earliest log.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 10:31 am
by Alan Kirk
John Hammond wrote:TM1 cannot find change I know I made in the log

Your start date was prior to the earliest existing log file. Change it to match the earliest log.
That, and/or the fact that the logs are normally stored in UTC time, not local time unless an undocumented config parameter is used. Searches therefore have to be done based on that time value rather than your local one.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 11:12 am
by lotsaram
Alan Kirk wrote:
John Hammond wrote:TM1 cannot find change I know I made in the log

Your start date was prior to the earliest existing log file. Change it to match the earliest log.
That, and/or the fact that the logs are normally stored in UTC time, not local time unless an undocumented config parameter is used. Searches therefore have to be done based on that time value rather than your local one.
Actually I'm not sure whether this is a bug or an irritant but even if tm1s-log.properties has the setting log4j.appender.S1.TimeZone=Local then this applies only to the tm1server.log (or the "message log" as opposed to the transaction log). This can be confusing when querying the transaction log as the admin must remember to offset local time to UMT when putting in query parameters. Thankfully it is not too much of an issue for us because CET is only +1 from UMT but I imagine this is much more of a pain if you are in a timezone with a much larger UMT offset.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 11:36 am
by Alan Kirk
lotsaram wrote:
Alan Kirk wrote:
John Hammond wrote:TM1 cannot find change I know I made in the log

Your start date was prior to the earliest existing log file. Change it to match the earliest log.
That, and/or the fact that the logs are normally stored in UTC time, not local time unless an undocumented config parameter is used. Searches therefore have to be done based on that time value rather than your local one.
Actually I'm not sure whether this is a bug or an irritant but even if tm1s-log.properties has the setting log4j.appender.S1.TimeZone=Local then this applies only to the tm1server.log (or the "message log" as opposed to the transaction log). This can be confusing when querying the transaction log as the admin must remember to offset local time to UMT when putting in query parameters. Thankfully it is not too much of an issue for us because CET is only +1 from UMT but I imagine this is much more of a pain if you are in a timezone with a much larger UMT offset.
Thanks Lotsa; I was deliberately vague on that point because (a) I've never used it and couldn't vouch for the extent to which it makes that change and (b) I'd suggest that most people don't use it anyway so even if it did affect the transaction log the overwhelming majority would still need to do a UTC query. Having transaction log files limited to UTC makes some sort of sense since they can be required for replication across time zones (when IBM doesn't screw up the format of the logs, anyway), and it also means no "doubled up hours" (no 2am to 3am followed by 2am to 3am again) and no missing hours. (However having a half way house situation where some logs can be local and some will always be UTC strikes me as being more confusing than even having to remember to convert to UTC. While I may not call "bug", I'd certainly call "badly thought out feature".)

Now UTC chore scheduling, on the other hand... :x

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 2:26 pm
by John Hammond
Thanks Lotsa / Alan

TM1editor - code goes to right of screen semicolon starts lines

This is most peculiar, and, if you have done a big edit, worrying. If you save and then reopen the display will revert to normal.
(If you are super cautious do 'Save As')

Anyone know what key combo triggers this?

edit look at this

http://www.tm1forum.com/viewtopic.php?f ... 212#p26212

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 07, 2011 6:40 pm
by Alan Kirk
John Hammond wrote: TM1editor - code goes to right of screen semicolon starts lines
Not following; can you describe this further?

Re: Cryptic TM1 Messages Decoded

Posted: Thu Sep 08, 2011 11:04 am
by John Hammond

Code: Select all

# Now set up the view
CubeName = CellGetS('Menu',Menu,'CubeName') ;
IF ( CubeName @<> '' ) ;
   ViewName = CellGetS('Menu',Menu,'ViewName') ;
On accidentally typing a key combination suddenly turns to:

Code: Select all

                                                                           # Now set up the view
                                     ;CubeName = CellGetS('Menu',Menu,'CubeName') 
                                                                         ; IF ( CubeName @<> '' ) 
                                     ; ViewName = CellGetS('Menu',Menu,'ViewName') 

Re: Cryptic TM1 Messages Decoded

Posted: Thu Sep 08, 2011 11:47 am
by Alan Kirk
John Hammond wrote:

Code: Select all

# Now set up the view
CubeName = CellGetS('Menu',Menu,'CubeName') ;
IF ( CubeName @<> '' ) ;
   ViewName = CellGetS('Menu',Menu,'ViewName') ;
On accidentally typing a key combination suddenly turns to:

Code: Select all

                                                                           # Now set up the view
                                     ;CubeName = CellGetS('Menu',Menu,'CubeName') 
                                                                         ; IF ( CubeName @<> '' ) 
                                     ; ViewName = CellGetS('Menu',Menu,'ViewName') 
I've never seen it happen. I've run through pretty much every combination of keys and modifier keys possible, and haven't found it happening on any of them. (Although that's in a 9.5.2 session; I'll take a look at 9.0 tomorrow if I have time.) Anyone else?

Re: Cryptic TM1 Messages Decoded

Posted: Thu Sep 08, 2011 8:09 pm
by Steve Rowe
the right hand ctrl+shift keys and the ctrl+shift left hand keys apear to have this affect in word and note pad, is it the same in the rule editor?

(Edit : retyped as I misread what the original short cut was)

Re: Cryptic TM1 Messages Decoded

Posted: Thu Sep 08, 2011 8:44 pm
by Alan Kirk
Steve Rowe wrote:the right hand ctrl+shift keys and the ctrl+shift left hand keys apear to have this affect in word and note pad, is it the same in the rule editor?

(Edit : retyped as I misread what the original short cut was)
I found a couple of posts that you alluded to in your original post, though I can't get those shortcuts to work in Wordpad or Notepad either. (Indeed the Notepad incarnation in Windows XP doesn't seem to have any option for right aligning at all. The one in Windows 7 has a "Right To Left Reading Order" option on the context menu, which will right align the text.)

Even Notepad is a sophisticated app compared to the TI editor, though, which consisted of someone at Applix dropping a text box onto a form in 1998 and Iboglix never lifting a finger to update the sodding thing since then because they didn't think that it would have enough of a "wow" factor for their sales presentations. :evil:

However... even if those shortcut keys worked (and they didn't when I tested them in 9.5.2 last night or 9.0 this morning), right alignment alone wouldn't explain how the code shifts to the right hand side of the semi-colon line terminators. Alignment is just formatting, this changes content as well. If John has posted exactly what has happened, that would scream "bug" to me though at this point it's a bug that I for one can't replicate.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 14, 2011 2:11 pm
by John Hammond
Simple TM1 Rules do not work
Simple TM1 Rules do not show on Rules Tracer
You are doubting your own sanity
You are doubt you have any understanding of TM1whatsoever

Solution

You have a second skipcheck section following the section where you have entered the rule which does not show up. The rules in the second area override the first.

This situation typically arises when you edit rules in an external editor and append the rules rather than overwriting them.

Re: Cryptic TM1 Messages Decoded

Posted: Wed Sep 28, 2011 5:59 pm
by John Hammond
TM1 Cannot open cube - cubename when you run chore or proc

You or someone on the server previously cancelled a proc with the cancel button on the progress bar. This seems sometimes to leave a lock on the cube but you cannot see this in TM1TOP.

Only solution I could find was to recycle the server after failing to solve the problem by cancelling ids in TM1top.

Not sure whether this might have been relevant as did not have a chance to check for cub$ files. Have a look at this

http://www.tm1forum.com/viewtopic.php?p=5791

Any more light on this one would be much appreciated - we now tell users to not cancel short running procs in the meantime...