..when the current job needs more space than you have.
This fixes #40.
The way it works is by having each datasource generate a list of its cached DEMs, we then subtract from this set the set required by the current job. What remains is a list of DEMs that we could remove if we need the space. We save this list in shared memory and in a synchronized manner, we delete these files when the unpack operation fails.
We still need to catch just the exception thrown by unpacking to a full disk though. I'm still testing this so its not quite ready yet. Just wanted to show the diff for easier debugging.
zerebubuth
commented
8 years ago
..when the current job needs more space than you have.
This fixes #40.
The way it works is by having each datasource generate a list of its cached DEMs, we then subtract from this set the set required by the current job. What remains is a list of DEMs that we could remove if we need the space. We save this list in shared memory and in a synchronized manner, we delete these files when the
unpackoperation fails.We still need to catch just the exception thrown by unpacking to a full disk though. I'm still testing this so its not quite ready yet. Just wanted to show the diff for easier debugging.