I'm getting the following error when processing an image dataset of several 100GB in KNIME Analytics Platform:
Execute failed: Error while writing to buffer, failed to write to file "knime_container_20160301_3895889446443796614.bin.gz": There is not enough space on the disk
I noticed that KNIME (on Windows) seems to write its buffer to the C: drive, which in my case is a limited-capacity SSD. I configured KNIME to use a workspace directory on another drive (D:), but how can I configure it to also save its buffer to another drive? Does it respect any environment variable that I could modify? I didn't find any documentation on that.
(Something similar was reported in this forum thread, but the solutions proposed to change the workflow rather than the configuration.)
Also, I suggest to Loop over your data if possible (e.g. by using the Chunk Loop Start / End pattern or with Simple Streaming). Like this, only the last node is persisted and not all the intermediate results, which you potentially don't need anymore.
File -> Preferences -> KNIME -> Directory for temporary files
Thanks, I somehow missed that configuration option.
I suggest to Loop over your data if possible
Yep, I doing this now. Actually, it would be great to be able to inspect the output of a loop end node (i.e. the partially filled table) while it is running. Is something like this possible?
Yep, I doing this now. Actually, it would be great to be able to inspect the output of a loop end node (i.e. the partially filled table) while it is running. Is something like this possible?
As far as I know it's on the list. Don't know when it will be implemented, though. I will point the developers on this topic. Thanks as usual for the valuable feedback :-)