Quick question - When mounting an rclone union remote with multiple back end remotes, and logging out to a file with -vv, if a log entry or error is generated that is specific to a particular remote, e.g. DownloadQuotaExceeded or Failed Authentication, is there any flag that will log the backend remote name in the log output as well?
Second, I can't really tell what the difference is when using all vs ff on the search policy. I've read the documentation and tried both mount settings, but don't see a difference in behavior, and am curious what is different with these two policies. When using ff for action or create - is it litterally the first found path in the order the upstreams are listed in the union, or is it the first remote to respond with the path?
e.g. if my upstreams are remote1: remote2: and both have /Files/, and I go to write /Files/FileA.bin, will ff pick remote1 everytime, or will it query both and if remote2 responds first, it will write there?
Sorry for all the questions, just love knowing the in and outs of this awesome tool so I can best utilize it and help others utilize it!
-Ed
Thanks in advance,
-Ed
Based on my own testing, with regards to search policy,
If you have a union with remoteA: and remoteB: as upstream, and content such as:
remoteA:/folder/file1.txt
remoteA:/folder/file2.txt
remoteB:/folder/file2.txt
remoteB:/folder/file3.txt
Regardless of what your search policy is set to, the union remote will show file1.txt, file2.txt and file3.txt in the /folder path
If you have the policy set to ff, it will use remoteA for file2.txt every time. If your policy is set to all or epall, it will query both remoteA and B and the first one to respond will present file2.txt.
So if all files are identical, ff seems more efficient as it yields less api hits.
For action or create, when set to all or epall, it will conduct those write actions against all upstream (assuming they are all RW for create tasks or at least NC for action tasks.)
If set to ff or epff, I believe it will conduct those actions on the first remote in the order of upstream in which the path exists. If none have the path then it will be the first remote with the appropriate RW or NC permissions.
I use ff on search_policy in a union off local disk and google cloud remote. The ideia was read only from local disk but is also downloading data from remote backend.