As many of you we have a million CI elements populated CMDB. When we try to generate some extractions that has lot of information it brokes (usuarlly by timeout). We have look up many alternatives and none fits at some size of extraction. We have tried:
1) split the tql in many to be downloadable.
2) user connect-it for download it through scenario.
3) develop java app to make the extraction.
Point 1) is not a solution for relally large downloads (would be more than 100 splits).
Poitn 2) generate stress to ucmdb server at some point we stop the scenario to no collapse ucmdb server.
Do you have any recipe for make large downloads easy? (not necessary fast, we can wait dyas).
Thanks in advance, Miquel
Pd: Kudos assured 🙂
There are a few setting that we can modify but it is not recomendable.
- Increasing Chunk Size
- Fuse settings
- Increase Java Heap Size
-Increase timeout fuse
I do recommend to verify that you are following the Sizing guide for your server and based on that change the settings mentioned.
If the above does not work, for further analysis, please raise a ticket with support.
Thanks for answering. I think we have tried all, but we gone a review.
Also we have some support interactions done, but we didn't find any good solution.
Can you give more information what do you try to extract? Is it a single class, or a complex TQL report?
Single Classes are the easiest, you can export only the attributes you need and you can extract quite big reports. When you start with related CIs, it is best to use the "Add distinct list" function in the reports, instead of exporting the relationships. This way you will end up with flat report which can have far more enries, than complex multi-layer report.
Thanks for answering :).
We, for example, are trying to extract computer and installed software, this take day's of tries and multiple files. Also for example we try to extract node and ip address. It also need like 3 different tql (split) to get the full content.
We don't know whats "Add distinct list" function, where/how we can manage/use it?
Thanks again, Miquel
When you go to the report tab of the view you can select the hierarchy and then Hide the IPAddress from the query:
However you need the IP address data for the Computers. You use the distinct function to add it back:
Once you have this, you can preview the report and decide if this is what you want. This way the report can become far longer than the previous TQl separation one: