osbi / saiku Goto Github PK

When publishing a new schema, JPivot view can see new cubes once the Mondrian Schema Cache is refreshed via the admin console (I haven't tested it in the Pentaho web front end) but the only way to get Saiku to see a new schema is to restart the Pentaho server.

Maybe I am doing something wrong but I can't fathom this one out?

I have an error when opening a query.
When I save the query, it generates an xml file with "UTF-8" encoding.
The MDX query contained in the xml file has special characters such as "á, ñ, etc.". So when I try to open the query, it results in an error.

I realized that when I change the xml encoding to "iso-8859-1", the MDX query is opened OK.
Is it a bug?

See here:

http://jira.pentaho.com/browse/BISERVER-6264

I'd just like to see the usual audit logs, i.e. who time and what.

"What" would be:

1. Launched Saiku
2. Ran Saiku query.
3. Saved Query
4. Loaded Query

That'd be cool. I Dont think anything more detailed than that is required, what do others think?

at the moment it is possible to select different levels from different hierarchies from a dimension
we should either disallow that or better, fix it in the backend so its ok to do so

-datasources
-refresh connection
-clear Mondrian cache

-etc

Allow for annotations similar to comments in Google Docs that allow for Dimensions, Levels, Members, and cell segments to be annotated and commented upon.

Add server side capabilities for addressing/reading/writing annotations
Come up with a limited syntax for comments (some easy to use Wiki Markup language seems like a good target)
Display the "availability" of comments on the pivot table, and a detail pane (to the right) that can show comments as a stream (or details if clicked from the pivot table)

Either return { "message": "OK" } or an empty body. The status code should be enough.

Saiku should be able to pass back the Locale to the Mondrian / XMLA Server when switching languages

When the user requests an XML endpoint with an Accept of application/json, the XML should be wrapped like this:

{ "xml": " ... " }

As far as I can tell, this affects:

GET /{user}/query/{query_name}/xml
GET /{user}/repository/{query_name}

When I export a saiku result set to excel, the numeric values in the .xls file are somehow stored incorrectly; excel says they are "number stored as text" and if I try to, say, do a sum(range) in excel, the answer is "zero".

Example excel export (with malfunctioning range sum) is attached.

Cheers,

Tyler

build seems to have troubles with maven3
especially with (http://download.java.net/maven/)

Can we filter on measures? i.e. where [Measures].[count] > 10 for example.

Commonly used for filtering out scenarios where there isnt enough data to make the averages meaningful..

Also availa

The children of the selected hierarchy elements are retrieved unnecessarily.

First reported in ##saiku and fixed by Paul:

(11:11:32 AM) cwarden: saiku seems to be too aggressive about descending hierarchies
(11:12:13 AM) cwarden: for example, if i have a date dimension with year, month, and day levels, and report on a measure by year, saiku queries for all of the months under each year.
(11:12:56 AM) cwarden: any way to disable this behavior?
(11:14:44 AM) pstoellberger: can you describe that in more detail?
(11:14:57 AM) pstoellberger: are you watching SQL? MDX?
(11:15:43 AM) cwarden: i'm watching the sql
(11:16:43 AM) pstoellberger: and you mean during the member cache population?
(11:17:29 AM) pstoellberger: we dont touch any lower levels in our code.... so this must be default mondrian behaviour
(11:19:50 AM) cwarden: i don't see the same behavior with jpivot.
(11:20:05 AM) pstoellberger: if you can show me some mondrian sql log about that i might be able to help you
(11:20:06 AM) cwarden: here's an example of the sql: http://pastebin.com/mQPSHtJv
(11:21:10 AM) pstoellberger: hm
(11:21:22 AM) pstoellberger: i can send you a new jar that you could try
(11:21:25 AM) pstoellberger: i think i have an idea
(11:21:46 AM) cwarden: cool, thanks.
(11:22:48 AM) pstoellberger: if its not that its not in our code base :)
(11:23:32 AM) pstoellberger: http://dl.dropbox.com/u/13927708/saiku-olap-2.2-SNAPSHOT.jar
(11:24:01 AM) pstoellberger: replace your current saiku-olap.jar with that one
(11:24:09 AM) pstoellberger: (make sure you delete/move the old one)
(11:24:24 AM) pstoellberger: tomcat/webapps/saiku/WEB-INF/lib
(11:24:57 AM) pstoellberger: or saiku/lib in the plugin
(11:38:54 AM) pstoellberger: any luck cwarden ?
(11:41:25 AM) cwarden: sorry, one sec. just being a little paranoid and checking the jar against the ci one. :)
(11:41:55 AM) thinkjson: it's a trojan, so he can look at what you're looking at
(11:41:59 AM) thinkjson: j/k
(11:42:00 AM) pstoellberger: lol
(11:42:06 AM) pstoellberger: i just commented about 10 lines of code
(11:42:21 AM) pstoellberger: you can do the changes yourself if you want :)
(11:42:37 AM) cwarden: CheatCellSetFormatter.java?
(11:43:41 AM) pstoellberger: yep
(11:44:16 AM) pstoellberger: line 535-551
(11:44:28 AM) pstoellberger: thats really paranoid :P
(11:44:56 AM) cwarden: half paranoia. half trying to learn a little about the architecture.
(11:45:07 AM) cwarden: anyway, you fixed it
(11:45:11 AM) cwarden: thanks!
(11:45:16 AM) pstoellberger: so it helped?
(11:45:26 AM) pstoellberger: hm yeah i knew that we do lots of stuff we dont have to
(11:45:47 AM) pstoellberger: in the code of those formatters might be dragons
(11:46:04 AM) pstoellberger: rewriting them is on my todo :)

http://analytical-labs.com/0001-Don-t-automatically-retrieve-children.patch

It would be helpful to be able to reload add/update/delete datasources on the fly. Without having to restart saiku-server. I also noticed that the plug-in has the same issue when running under pentaho.

Thanks
Eric Smith

In the output table the columns order doesn't follow what specified in the columns definition.
If you look at the attached image you can clearly see that the Units ordered measure must precede the
Units shipped measure but in the output table it doesn't happen.

The restful resource BasicRepositoryResource performs saves of queries to files directly, it does not delegate off to another class. In order to embed saiku in existing applications it would be benefical to have some sort of query persistence service which it delegates to and would allow for alternative implementations to be written and plugged in (db persistance for example).

When you try to select the members on a dimension in a query the selection is not taken into account.

Pentaho BI server 3.8 + Saiku Plugin + Firefox 6.0.2 (also IE 8 )

the current layout makes the width of the list too small, so you cant really properly read the REST urls

for enunciate servlet

Create and enable an optional query param for many metadata and query dataset REST endpoints that allow for showing properties or not.

Enable all the properties sections (that have otherwise been disabled/commented out) along with modify constructors of objects that get json-i-fied to ensure they aren't output if not asked for specifically and are output if asked for specifically.

Default should be for off, queryparam name should be
showProps = true/false

additionally to the datasources.xml you should be able to define custom datasources, just like in the standalone version

I'm not sure if you can put textual help info into the schema, but if you can it would be nice to display this when the cube is selected.

In fact it would be nice if you could do this in some wiki style way. Allow users to annotate saved queries, and indeed cubes themselves! :)

Drive that on further, allow saving of resultsets, and you can annotate those too! Wouldnt that be cool....

Currently, all the JSON returns are List<> objects and this can, in some circumstances, be a security risk. See http://flask.pocoo.org/docs/security/#json-security.

Additionally, Enunciate does not document datatypes properly anyhow so it's good to do this as well.

Add ability to query and see per group totals. Basically, in addition to displaying just the lowest level cells, retrieve the totals for other levels of granularity and display those.

The /mdx endpoint sends application/json Content-type but returns text. Backbone can't read the returned data.

Request:
Host saiku-ui.tiemonster.cloud9ide.com
User-Agent Mozilla/5.0 (Windows NT 6.1; WOW64; rv:5.0) Gecko/20100101 Firefox/5.0
Accept application/json, text/javascript, /*; q=0.01
Accept-Language en-us,en;q=0.5
Accept-Encoding gzip, deflate
Accept-Charset ISO-8859-1,utf-8;q=0.7,;q=0.7
Connection keep-alive
X-Requested-With XMLHttpRequest
Referer http://saiku-ui.tiemonster.cloud9ide.com/

Expected response:
Date Wed, 27 Jul 2011 17:33:34 GMT
Content-Type application/json
Vary Accept-Encoding
Content-Encoding gzip
Content-Length 124
Keep-Alive timeout=15, max=91
Connection Keep-Alive

{ mdx: "SELECT
NON EMPTY {Hierarchize({[Gender].[Gender].Members})} ON COLUMNS,
NON EMPTY {Hierarchize({[Product].[Product Family].Members})} ON ROWS
FROM [Sales 2]" }

Returned response:
Date Wed, 27 Jul 2011 17:33:34 GMT
Content-Type application/json
Vary Accept-Encoding
Content-Encoding gzip
Content-Length 124
Keep-Alive timeout=15, max=91
Connection Keep-Alive

SELECT
NON EMPTY {Hierarchize({[Gender].[Gender].Members})} ON COLUMNS,
NON EMPTY {Hierarchize({[Product].[Product Family].Members})} ON ROWS
FROM [Sales 2]

Hello,
I'm having some problems when I select a member in a parent-child hierarchie with some ancestors.
For example:
If I build by my own query: "SELECT
NON EMPTY {Hierarchize({[Dim1].[member1]})} ON COLUMNS,
NON EMPTY {Hierarchize({{[Dim2].[All Dim2]},{[Parent1].[Child1]},{[Parent1].[Child1].[Child2]},{[Parent1].[Child1].[Child2].[Child3]}})} ON ROWS
FROM [Horas]"
this works perfectly and shows all hierarchies with all Totals.

But if I try to do this in a new query in saiku, selecting in columns member1 and in rows Child3 respectively, this throws an exception like this:

"org.olap4j.OlapException: Unable to find a member with name [[Parent1],[Child2],[Child3]]"
As you can see, it try to build a query skipping some members in the hierarchie.

I guess that this is a bug.

mondrian 3.3 explain plan

many nodes, same cache, for scaling etc

Calculated members and calculated member properties that are part of just the MDX are useful for both user facing features (YTD totals) but also are an enabling piece to create a UI to do conditional formatting, outside drilling, etc.

Add ability to Saiku (and potential OLAP4J since I just looked and it doesn't appear the query model currently has the ability to do this) to create per query Calculated Members and Calculated Member properties.

I have a virtual cube definition with calculated members. These are not displayed in the query's measures list.

see brief discussion in IRC:

codek: for the time being, you could switch of caching at schema level
it had dawned on me that we have no way to flush the cache at the mo

cool, we need that admin console asap ;)
what we should create is a rest endpoint for cache clearing, so when your ETL is complete it can hit a Saiku URL and flush the cache for you

Hi,

is it possible to make the Saiku BI Server plugin to pick up the standard mondrian.properties file that is located in pentaho-solutions/system/mondrian.

Otherwise you have to keep multiple mondrian.properties in sync.

Thx,
Ingo

http://admin:[email protected]/saiku/rest/saiku/admin/discover/foodmart/FoodMart/FoodMart/Sales%202/dimensions?_=1311709976233

http://admin:[email protected]/saiku/rest/saiku/admin/discover/foodmart/FoodMart/FoodMart/Sales%202/measures?_=1311709976414

The probleme is when you save a query with a login like admin, a
different

person with diferrent login can acces to the query who were saved by admin login

When using Pentaho permissions for schemas (one different schema for each user), all the users can see all the schema inside Saiku plugin. The same do no happen in Pentaho's Analysis View.

This is an amazing tool, and I am tremendously excited to use it. I'm having some trouble getting everything hooked up, though. I'm running Mondrian, Saiku, and Saiku-ui on a Tomcat server on Ubuntu, with a Mysql database storing the data. I have set up a datasource/catalog for Mondrian, and can access basic MDX queries through its web interface. I have tried to set up Saiku to use the same data source (following instructions at http://projects.analytical-labs.com/knowledgebase/articles/3, adding a file to WEB-INF/classes/saiku-datasources), but am not sure if I have set it up correctly or not, because when I access the Saiku-ui, there are no Cubes available in the drop-down.

I get the following information from /saiku/rest/saiku/admin/datasources:

[{"name":"Users","properties":{"location":"jdbc:mondrian:Jdbc=jdbc:mysql://localhost/grockit?user=root&password=password;JdbcUser=root;JdbcPassword=password;JdbcDrivers=com.mysql.jdbc.Driver;Catalog=/WEB-INF/queries/Users.xml","password":"password","name":"Users","type":"OLAP","driver":"mondrian.olap4j.MondrianOlap4jDriver","username":"root"},"type":"OLAP"}]

Is there a good place to look for logs to help find debugging information, or even just to know whether or not the saiku server has successfully processed and accessed the data source?

Many thanks -- I'm very excited about using Saiku!

While the message is "INTERNAL_SERVER_ERROR" and the JSON doesn't parse, the HTTP status code is still 200. This fires the success callback in jQuery.

the filter axis is using hierarchyConsistent and crossjoins for mdx
needs to be reviewed

I would like to pass parameters in the request URL to be included in the MDX query. I'm not sure what the interface looks like, but conceptually, I'd like to be able to do something like /saiku/rest/saiku/user/query/(id)/result?gender=male and /saiku/rest/saiku/user/query/(id)/result?gender=female.

Instead of the properties property on the properties endpoint, this endpoint should accept arbitrary key-value pairs in the body itself

For example:

Old request
properties saiku.olap.query.automatic_execution false
saiku.olap.query.nonempty true
saiku.olap.query.nonempty.rows true
saiku.olap.query.nonempty.columns true
saiku.olap.query.drillthrough true

New request
saiku.olap.query.automatic_execution false
saiku.olap.query.nonempty true
saiku.olap.query.nonempty.rows true
saiku.olap.query.nonempty.columns true
saiku.olap.query.drillthrough true

Add a warning when the number of rows is too high, maybe with an option to cancel the display of the data.

Add an option to configure/change that threshold.

The version of the MySQL JDBC driver bundled with Saiku ignores the TCP port in the connection string and always connects to port 3306. Replacing the driver with 5.1.17 fixes the problem.

...instead of clobbering

we need 1 zip with all sources as download
the source version of the jars in an artifactory

see olap4j test code:

final OlapStatement olapStatement = olapConnection.createStatement();
final Throwable[] exceptions = {null};
new Thread(
new Runnable() {
public void run() {
try {
Thread.sleep(1000);
olapStatement.cancel();
} catch (Throwable e) {
exceptions[0] = e;
}
}
}
).start();

use a backend call to check if the credentials are alright or not
if not -> don't login

Dynamic Schema Processors are powerfull means for customization. i.e. I18N, multi tenancy or dynamic roles.
Please make the Saiku Plugin aware of these processors.

http://mondrian.pentaho.com/documentation/schema.php#Schema_processor

access a query via any of the endpoints (csv, json, excel) that throws an OLAP4J exception (invalid query, etc)

Should propagate the error to the client (40X error or JSON document with error information, or Redirect to JSON with error, etc).