Parallel file hashing in 2.5.x

I’ve already used Parallel file hashing on client successfully (= no errors) with urbackup 2.4.x.
Now as it seems in Urbackup 2.5.x there are a few new settings for it, which for me raises a few questions:

Especially the setting: “Beta: Number of parallel client file hash threads per file backup” which by default is set to 1.

  • If I activate the Parallel client file hashing for the internet clients, but the allowed threads to use is 1, wouldn’t that effectively disable parallel file hashing?
  • Why can we only activate parallel file hashing for internet clients? wouldn’t it also speed up the indexing for local clients?
  • Or when exactly are the files being hashed, during indexing? Or during the transfer phase, and parallel file hashing just means, that while urbackup uploads a chunk, it will parallelly hash the next chunk(s)?

I can’t find any information whatsoever about parallel file hashing in the admin documentation.

I’m also curious about this, as I’m trying to back up a server with 13TB of data in 6 million files, which took five days to do an initial backup. It’s now several hours into an incremental backup which should only be a few GB difference, and according to the live log is about 1.6 million files in:

02/13/23 11:41 DEBUG Skip phash for id 1682916 2eO/VVc8j8yXKj+VW/m2VykaRJwyn+/48MuBCbB3eQBjkmGfPUHV3z2rmRcUZB5eJRX+yK6um3fPEfWkTmeT7A–

Might increasing the file hash threads here speed up this “skip phash” phase?

This means it is receiving a hash that it didn’t need. The question would be more why this hash wasn’t stored to the database and then re-used…

W.r.t. the original question. The parallel hashing concerns (from the manual):

If checked the client calculates hashes for each file before the backups (only hashes of changed files are calculated). The file then does not have to be transferred if another client already transferred the same file

Without parallel hashing this is done during indexing. With it is done during backup (in parallel), with one thread. The options are to increase the number of threads, but I wouldn’t touch them currently in production.

Ah, interesting. I’ll restart things in debug mode on both sides and see if I can get more information, and start a new topic.