If curl download times out, curl isn't terminated, partial download isn't removed & next run fails.

[dlmetcalf]

If the curl download of confluence tgz isn't successful, it appears that any partially downloaded file isn't removed and the curl background process isn't killed. (This is probably a Staging module bug, rather than Confluence module bug).

When reattempting puppet apply, instead of reattempting the download, it sees the partially downloaded (corrupt/incomplete) file and starts extracting it before failing.

If the files are removed (e.g. rm -r /opt/confluence /opt/staging/confluence) and puppet re-run; multiple curl processes then appear to be running.

[igalic]

i'm pretty and sure we fixed this with an upgrade in archive… which now may be causing different problems, but this one is fixed :D D: