Monthly Archives: September 2007

How to troubleshoot mysterious SharePoint errors.

Overview:

Debugging is difficult when developing custom functionality for Windows SharePoint Services 3.0 (WSS) or Microsoft Office SharePoint Server (MOSS).  The main culprit is that SharePoint normally surfaces very little diagnostic information on the web browser when an error occurs.  This blog entry describes how to locate additional system-generated diagnostic information that can often provide that extra bit of detail that one needs in order to identify root causes.  This can then lead to solving the problem.

I have used this technique with great success to solve otherwise mysterious errors.

Approach:

SharePoint saves a great deal of information to a diagnostic log in a log file in the 12 hive.

The "12 hive" is usually located at "C:\Program Files\Common Files\Microsoft Shared\Web Server Extensions\12".  (I’m not sure if it’s possible for the 12 hive to live anywhere else, in fact).

The idea is to locate the current log file, force the error and then quickly open the log file.  These log files are characterized by:

  • Copious amounts of information.  SharePoint generates a very large amount of diagnostic information and writes it to that log file very quickly.  You need to be quick with the fingers to capture it.
  • Multiplicity.  SharePoint does not write to a single log file but rather generates multiple log files in sequence. 
  • Copy and paste nicely into MS Excel.

My favorite method:

  1. Open up a windows explorer pointing to the 12 hive\logs.
  2. Sort the view to show by modified date (most recent first).
  3. Highlight the most current log file.
  4. In a web browser window, force the error to occur.
  5. Quickly open the current log file and copy its contents to MS Excel.
  6. Jump to the end and analyze the relevant entries.

Other Notes:

By default, the diagnostic log is located in the 12 hive\LOGS directory.

MS Best practices (as per Mike T. of Microsoft) state that the log files should be saved to a separate hard drive.  One does this via central admin.  Your system administrator may have done this, in which case you’d obviously need to find the log file there instead of the default 12 hive location).

This entry addresses issues such as:

  • SharePoint workflow failed to start due to an internal error.
  • (more to be added over time)
  • This entry has been helpful diagnosing workflow errors (e.g. "The workflow failed to start due to an internal error").

 

 

MOSS: Effective introduction to an Organization

(this entry cross posted between http://paulgalvin.spaces.live.com/blog/ and http://blogs.conchango.com)

The postings on this site are my own and don’t necessarily represent Conchango’s positions, strategies or opinions.

Overview

This entry describes some background information on a large (3,000 users) Microsoft Office SharePoint Server (MOSS) rollout and what we did to get the project rolling in such a way that the client is happy and firmly down a path that ends with full adoption of the MOSS feature set.  As of the writing of the entry, we are approximately 50% complete with the first phase of the project.  As things progress, I’ll update this entry and/or write new entries.

In this specific case, the company had already installed SharePoint Portal Server 2003.  The IT group installed the product in a sort of "let’s see if anyone cares" fashion.  It was quickly adopted by many business users and became quite popular in the enterprise at large.  As you can imagine, this was not the best rollout strategy (which the client readily admits) and when MOSS arrived on the scene, the client resolved to "do it right" and hired us to help them.

One of the central questions facing us when we started implementing this project was: How do we introduce MOSS to this client?  Given that the client already had experience with SharePoint, we wondered — do we need to do "differential" training or do we start from the ground up?  After working with key users, we determined that treating this as a green field project made more sense. 

That decision gave a starting point but still left us with the major requirement of figuring out a good strategy for rolling MOSS out to the enterprise.  MOSS is such a big animal … it includes content management, document management, search, security, audience targeting, project management, "fabulous forty" templates, workflow, business data connector, etc.  Couple this with the fact that it’s a large organization that can truly make use of virtually every major MOSS feature and you have the makings of a great project with an enterprise reach and many good things happening.

We’re confronted with this issue time and time again … MOSS has an enterprise reach with its enterprise feature-set, yet even somewhat sophisticated clients have a hard time mentally absorbing those features, let alone incorporating an appreciable fraction of them into their daily routine.

I don’t have a magic solution to the problem.  I instead address just the very first steps that we’ve taken with the client to lead them down the path to successful long-term adoption.

Scope:

As much as I wanted the team to craft a project plan that included such milestones as "PoepleSoft Integration via BDC Completed", "New Cross-Departmental Product Launch Workflow Complete" and "Executive Management KPI’s Accepted", I had to settle for something less.  This is not to say that "less" is bad. In fact, the "less" that we decided for the initial rollout was miles ahead of where they were before we started.  In our case, the "less" turned into:

  • Simple document management using document libraries, version control and content types.
  • Effective search based on content types and customized advance search (via managed properties, XSLT to produce pretty results, etc).

In addition to the above enterprise-wide features (meaning that they were to be rolled out to all departments and users), we added the following singleton in-scope mini-projects:

  • Proof of concept BDC integration.
  • Multi-step and multi-branch workflow process created via SPD.
  • Complex InfoPath form.
  • Surfacing KPI’s for some business process (probably HR talent acquisition in our case, though that may change).

The scope here is not 100% accurate but representative of our approach and sufficient for my purpose here, which is to explain what I consider to be an "effective" introduction of MOSS that will set the client firmly down the golden path to full MOSS adoption.

I won’t write much more about the singleton in this entry.  I do want to point out that these are part of our over-arching strategy.  The idea is to implement the core document management and search features to all users yet provide highly functional, high visible and highly representative examples of other core MOSS features which are simply beyond the ability of most users to absorb at this early stage.  However, they will be "out there" and one hopes that other business units will know of or learn about them and want those features for themselves, leading to greater adoption.  These singleton success stories also serve to provide our sales team "ammunition" for successfully winning second, third and n-phase projects.

What Did We Introduce and Why? 

Having settled on document management and search as a baseline enterprise-wide requirement, we needed to start gathering details.  As a practical matter, this revolved around understanding their documents and that ultimately mapped to understanding content types.

I’ve found it’s difficult to explain content types without visual aides.  More technical folk can walk away from a discussion about content types when CT’s are described in database terms.  "A CT is similar to a database table, it has columns and columns are defined in terms of data types, but CT data types include more than simple integer/date, but also "choice" and "lookup" and the like."  We can talk about "extending" content types, much like one can inherit functionality from a base class in object oriented languages.  However this is obviously not helpful for the transportation department admin person who has no technical background.  I.e., nearly everyone that matters in a MOSS rollout.

Using a white board is iffy.  I’ve presented the idea of a content type and drawn brilliant (or so they seem) pictures of content types and what they do for you in terms of search and how they can be expanded, etc.  In the end, it does feel like some light bulbs have turned on, but the resulting white board picture is a mess. 

This led us to our current and so far most effective landing place: a MOSS sandbox site configured to show these features.

Using the sandbox site, we demonstrate:

  • Content types:
    • Creating a CT with multiple data types (text, date, choice, boolean, lookup, etc).
    • Extending a CT by creating a new CT based on a parent.
    • Searching for documents using CT metadata.
  • Document libraries:
    • Associating a single CT with a library.
    • What happens when we upload a document to that library?
    • Associating multiple CT’s with a doc library.
    • What happens when we upload a document to that library?
    • Filtering and sorting via column headings in a doc lib.
    • Document library views:
      • Sorting
      • Grouping
      • "Quick entry" (data sheet view)
      • "Untagged data" (to assist with migration to MOSS from other content sources; more on this below).

The Sandbox Site:

We designed our sandbox site to be a permanent feature in the development environment to be used for training purposes long after we finish the project and included several artifacts as described:

Content Types:

We defined the following content types: Invoice, Purchase Order, Services Invoice.

We selected Invoice and Purchase order because they are more or less universally under
stood entities.  Everyone in business understands that an invoice is a demand for payment to a customer for an amount issued on a certain date to be paid as per some payment terms.  This leads to a natural definition of a CT which we called "Training Invoice" (to distinguish it from any other kind of invoice).  The purchase order is similarly easily defined.  We also created a "Training Services Invoice" by creating a new CT based upon the "Training Invoice" CT and added just one column, "services rendered".

With the above, we can now demonstrate some key features of CT’s without getting bogged down trying to explain an abstract concept first; everyone already understands what we mean by "invoice" and "purchase order" and are instead able to focus on mechanics of the CT itself.

Custom Lists:

CT with columns of type "lookup" point to a custom list or document library.  We use this extensively and for the sandbox, we created one supporting custom list that contains customers.  We picked customers because it’s an easy concept to understand and easy to demonstrate.  The Invoice CT has a column, "customer" that is defined of type "lookup" that points at this list. 

We created a similar custom list to manage "vendors" for the "Purchase Order" CT.

Document libraries:

We created two document libraries: "Invoices" and "Mixed Documents".

We configured the invoices document library to manage only documents of CT type "Invoice".

We configured the "Mixed Documents" library to manage all three CT’s.

Create several views that show sorting, filtering, data sheet and grouping.

Search:

We defined two new managed properties and mapped them to Invoice Number and Customer.

We created a new customized advance search site and modified it to enable users to search for "invoices" using those two mapped properties.

Modify the XSLT so that the invoice and customer number, when present, appear in an HTML table in a bright color.  The objective here is to demonstrate that such formatting is possible.

Putting it all together:

We arrange for key users to participate in a demo.

We follow this simple script:

  1. Describe the meaning and purpose of a CT, using invoices and purchase orders as examples.
  2. Show the invoice CT definition while simultaneously assuring them that they don’t need to use those screens themselves, just pick up the concepts.
  3. Go to the invoices document library.
  4. Upload a document.
  5. Demonstrate that the customer drop-down is really sourced from a custom list.
  6. Add a new customer to the customer list and then update the recently uploaded invoice’s meta data with the newly created customer.
  7. Switch to the "mixed documents" library and upload a document.  Explain how the system prompts for a document type.
  8. Go back to invoices document library and show how clicking on a column name changes sort order.
  9. Demonstrate column-level filtering.
  10. Show different views that demonstrate multi-level sorting, filtering and grouping.
  11. Show the data sheet view.
  12. Explain the purpose of a "untagged documents" view.
  13. Switch to the customized advanced search.
  14. By now, the recently uploaded document should have been crawled and indexed, so perform a search that demonstrates ability to locate that invoice via the mapped property.
  15. We demonstrate the difference between searching via mapped properties vs. just a text search.

At this point, we are more or less done with the demo.  It seems to take about 30 to 45 minutes, depending on how many questions people ask.

We then send them back to their desks with "homework".  This consists of a simple excel spreadsheet where we ask them to define for us what they think they need in terms of CT’s, both at a high level (just name and business purpose) as well as columns and type of data they would store in the column.  We don’t ask them to define column data types in MOSS terms, but business terms. 

In Summary:

We’ve created a sandbox environment that we can use to demonstrate some core MOSS features whose appeal are enterprise-wide. 

We have modeled easily understood and common business entities so that users can focus on MOSS and not get bogged down on the entities / examples themselves.

Business users walk away from theses sessions with "homework" in the form of excel documents which they are now competent to fill out and use for designing their own first-cut content types.

Finally, as we perform demos over time, the client’s team members themselves become more able to carry forward, do the demo’s themselves and generally free up the rest of us up to work on more complex issues, such as global taxonomy, complex workflows, BDC and the like.

Learning Agile // Scrum

I asked for some advice today on good resources for getting started with learning Agile and Scrum.  Here is a summary of the response.  I trust the sources but I don’t know that this is comprehensive (I’m sure it is not).

I may have transcribed some of this incorrectly.

Multiple people provided responses and Agile Project Management by Ken Scwaber is the consistent "first dive" recommendation.

Personalities:

  • Ken Schwaber
  • Mike Cohn

Books:

  • Agile Project Management with Scrum by Ken Schwaber.
  • Lean Software Development: An Agile Toolkit for Software Development Managers by Mary and Tom Poppendieck.
  • "anything by Mike Cohn"
  • Agile Retrospectives by Ken Schwaber, Diana Larsen, Esther Derby.

Links: