Monday, July 23, 2012

Transfer Order

If you had an experience of dealing with Maximo 7.5, especially in the part of Transfer Order or Internal Purchase Order, you will notice that Maximo is so freakin slow. To transfer about 1500 assets, you gonna ends up with a hung machine. The figure may be too much for Maximo to transfer. In Maximo, there are 2 activities of Transfer, we call it Transfer Out and Transfer In. We can breakdown the activities of both, to be more detail, but i dont want to share it out here. What i want to share out is that, the slowness of Maximo. I implemented Maximo in AORA, and you know the number of assets to transfer daily can reach about 1500 even more. All the people really got screw up with the slowness. Some more the websphere was restarting by itself because of lacking memory. Digging into the source of Maximo, makes me understand that there is a  Maximo developer who fucks up with the code.

Knowing that, I just modified the Transfer Order process. Streamlining the process as to use excel to perform Transfer Order and let the background process deal with the file will help users' tasks on using Maximo. Besides, having API of Maximo is really helpful in designing your code or doing customization of Maximo. And now, there is no more people of AORA got screwed because of the new design of Transfer Order.

One of the mechanism of Transfer Order that I have changed is related to the way Maximo creates record of invuselinesplit, shipment, shipmentline and matrectrans. Most of you may know what all of these tables for when you perform Transfer Out(Most of these tables are introduced in latest Maximo 7.5), these tables are used most. Maximo 7.5 process the records on the last 3 tables in the same method, which is during status change. Imagine, that you are going to Transfer Out a thousands of records and with the current process, Maximo will populate a thousands and more record in memory. That's y, Transfer Out process takes too much resources and takes too much time for user to complete. I changed the way this Transfer Out process does. So, When user will Transfer Out a thousand of records, the process will populate in the memory for about 20 records, after that, it will save it to DB. In this way, there will no more Out of Memory problem in your Maximo application when you do Transfer Out.

table or view does not exist

Just to share out something about Oracle during implementation of Maximo in my client. I was doing some integration between Maximo and MQ. As there is no clear way of doing integration between Maximo and MQ, we decided to use DB as the endpoint of Maximo integration to MQ. The data is only one way, that is from MQ to Maximo.We use Oracle package which runs every 8 am and 5 pm every day.

The things came up when I created stored procedure from other schema. If we test the query in the schema (not in stored procedure), it was smoothly being executed, but if we apply it in the stored procedure. The stored procedure wont be able to be compiled with error of nonexistence of table. Granting privilege will solve the problem

Sunday, October 30, 2011

Update field automaticaly once another field get updated

Now, I just want to share out a bit of how to update the field automatically in Maximo. A lot of time, we have requirement on how to update the field once another field get updated. Yes, this is really to ease user for not updating another field manually while the information exists in another table. Well, this kind of behaviour is quite common in the UI of maximo. Let's say in the Purchase Order application. Once you update the field of quantity, you should get the total cost of the purchase. that is, one of the default behavior of the Maximo UI.


In order to perform the above behavior, you can do it in 2 ways.
  1. Coding in the level of Mbo
  2. Coding in the level of Bean
Both must be done by coding. :P Well, it's indeed this blog is meant to be.
Anyway, doing coding in level Mbo, you can do it in Fld classes. Thus, you need to perform some changes on metada of Maximo's database. But, one thing that you need to ensure is that you update the same Mbo as what the bean is actually display. Because, a lot of time, the mbo object is not the same as what the mbo object used in Bean. Thus, updating the data will take no effect on the mbo which current bean displays.

doing coding in the level of Bean, will definitely give you the right Mbo. Thus, there is no need to worry about different Mbo object problem as I describe above.

Saturday, October 22, 2011

Maximo does not publish the changes created by MIF

During my project in AORA, I have been configuring a lot about integration/MIF as it deals a lot with external application so called Satellite Application. The idea is the satellite application does not deal directly to Maximo database, but they are accessing Maximo database through iface table. Well, that is general idea of why Integration exists.

In the business process of our client, there is a process to transfer their assets from one storeroom to another storeroom. Well, the satellite application will expect to have a record of Transfer Order to perform Transfer Out-In process. The record is created in the iface table (this is the term representing a table used to exchange data between external application and maximo). Transfer out will be performed first, from Maximo. Once, the process of transfer out is finished. Transfer in will be processed. Transfer in will not be executed if the transfer out has not been done. The transfer asset process involves a hundred of assets to be transferred even thousand, thus, we cannot use a standard Maximo to do this.

To fulfill the above scenario, I make some customization to process the assets. I am using excel reader to read excel and then process it to MIF. Using MIF, I am able to create record. Well, upon creating the record I am expecting to get a record for transfer in. But in fact, it does not get created which is weird. I have created a publish channel to monitor the objects but still the transfer in object does not get created. Well, my hypothesis is that it has different behavior between integration and created directly from UI, which I think, it is not supposed to be so. Thus, the solution to deal with with the problem is to perform some changes on the object either using mbo changes or using escalation to monitor the existence of object and do some changes (adding description to object or any kind of changes to trigger MIF to publish the changes) on that.