Can you help me understand what prompts the tmp files used by the HXTT Text (CSV) file to go away? We are using the 4.2 version and we have a case where the driver is creating about 3GB of tmp files. We finish our activity with the data and the files don't seem to be going away. Our issue is trying to figure out how big we will need to make this tmp space as we keep running out of it.
What's the suffix of those temporary files?
They are all .tmp files and there are quite a few of them. Below is the contents of the /tmp directory following our processing:
>We finish our activity with the data and the files don't seem to be going away.
1. When Java VM exit, it will be deleted.
2. HXTT will remove all older temporary files after about 6 hours according to one counter. It seems that frequency is too low. The latest version will remove temproary files after about 30 minutes.
Yikes. This driver is running inside the context of the application server so the VM really isn't going to exit. I'm not sure what the relationship is between these files and the JDBC connections or the "tables" being used in the quer(ies) but in our use cases we don't want these files laying around. We are really using the JDBC driver to import csv data so once we collect it we really don't have to go back to it. Ideally if these files are somehow related to the JDBC connection then after the connection is closed then it would be nice to see these cleaned up. I can understand why that might not make sense generally as I presume you might use these files if I open another connection to the same files in the future? But perhaps even a configurable parameter which tells the driver after how many seconds to clean up un-used temp files.
Lastly, when you say "the latest version" do you mean the version posted up on the site now, the next released version? We recently downloaded an update but if you could tell me the specific version that might at least give us a bit better relief here.
thanks! I appreciate the responsiveness of your support.
>the next released version?
Upload now. Please download it.
Thank you for your continued support. We are also noticing that in the 7/21/2010 released version of the driver, that the .TMP files will fill our /tmp directory in a very short amount of time. The /tmp directory has 1 to 3 GB of space, yet it fills to 100% capacity within seconds.
We did not notice this with previous versions of the driver.
Is there an issue with this latest version? The issue only seems to manifest itself with large data sets (ie: a .csv file with 500,000 rows, for example).
The SELECT queries contain 'DISTINCT' statements, as well as 'CASE' statements.
Is there a recommended amount of space to allocate to /tmp, for example?
Supported. PLease download the latest package.