Gdrive mounted on a ftp server

[BoNx]

Oh i forgot to mention... i dont use rclone for ftp. i already have my ftp server installed. so i use rclone only for mount my gdrive. I don't know if it makes a difference. And thx a lot for your help very appreciate

[asdffdsa (jojothehumanmonkey)]

gdrive supports polling.

[Humanplus]

Tell me if I am wrong, but --dir-cache-time=192h means that the cached/scanned directories are cached and not rescanned again (instead every X mins), changes are of course noticed.

The polling option sets how often it will be rescanned.

If I am wrong, then I can learn from it

But I see really better performance and a low # of api calls.

I have 20x Remotes à 5TB (= 100TB) -- wrapped in a Union --- and active encryption + encrypted folders & filenames.

[asdffdsa (jojothehumanmonkey)]

hi, not at expert at polling.

@Humanplus

changes are of course noticed

only if rclone makes the change
if a change happens in gdrive, that will not get noticed until the --poll-interval interval expires.

@Humanplus

The polling option sets how often it will be re-scanned.

as i understand it, the remote is not scanned. individual changes are noticed.

a re-scan only happens when

• --dir-cache-time interval expires.
  or
• on demand using rc vfs/refresh

[Humanplus]

@asdffdsa

only if rclone makes the change
if a change happens in gdrive, that will not get noticed until the --poll-interval interval expires.

Yes, asolutely - your are right.

But a two-way data exchange is much more complicated... wow, yes.

In this case I didn't mentioned that I meant just a ---one way --->

Root --> rclone --> remotes (GDrive/OneDrive/...) ---> Union --> Encryption --> /mnt/cloud --> SeaFile

And there is no problem with dir-depth/filelengh because SeaFile saves data in data blocks.

@asdffdsa

as i understand it, the remote is not scanned. individual changes are noticed.

a re-scan only happens when

• --dir-cache-time interval expires.
  or
• on demand using rc vfs/refresh

Interesting thesis ...

... I really can't tell you - and storage-cloud, big data is really not my department

But would't it be the best way to "recognise changes", catalog them, creating a kind of "map" and then to save this (data-)map to minimize the latency?

It would be really interesting to hear what the developer say.

[asdffdsa (jojothehumanmonkey)]

@Humanplus

But would't it be the best way to "recognise changes", catalog them, creating a kind of "map" and then to save this (data-)map to minimize the latency?

i think that is what rclone does, via polling

[Humanplus]

Good team-work Yup, this makes sense.