TheSeven

I just had another look and couldn't reproduce it with a simple test case either. However, it seems like the the PC that you create a folder on doesn't set that to fully synced automatically, which might have confused me, because there was basically a full copy of everything created before a certain date on that one. I'll try to observe this closer if it hits me again. Is there a way to check what Sync considers to be a fully synced subtree (that new files will automatically be pulled into) and what it doesn't?

I'm running Sync 2.4.4 (732) on Windows, not installed as a service. I also have another instance on Linux which I'm using through the WebUI, but I'm not using Selective Sync on that (i'm using it as the always-on "master server" for all my synced folders). I have a synced folder D:\Projects with Selective Sync enabled (on multiple Windows systems), and synced a different subset of subfolders (but having several in common) to these systems. Now assume that I have a folder D:\Projects\ProjectA synced to all of them, and I create a new file D:\Projects\ProjectA\Testfile.txt on one system. It will get synced to the "master server" (without Selective Sync) just fine, but on the other Windows systems (with Selective Sync), only a D:\Projects\ProjectA\Testfile.txt.rsls file will be created. As these folders contain source code where files are added regularly, and I want to be able to build from that subtree on any of the systems it's synced to, I have to right click the folder and select "sync to this PC" (or the translation of that in my OS language) every time I have added a file somewhere else, or the builds will fail due to missing files.

Actually the "Scenario 2" problem exists even in Windows Explorer. It isn't quite as bad because you can just click "sync folder to this device" on FooSubfolder to get all newly-created files, but it's annoying that I have to do that all the time, especially if I only realize that something is missing while I have no (or bad) internet connectivity. The reason why I have selective sync enabled is that I only want to sync some subtrees to that device. However I want these subtrees to be complete. It would be vastly preferable if any newly created files inside a fully-synced subtree (of a selective sync top-level folder) would automatically be downloaded as well. I could also live with an option to disable selective sync on certain subdirectories of a synced folder.

There has been further discussion with some more detail in support ticket #52811 if you want to take a look at that. I do understand why some things need to be flushed to disk, but if my understanding of this matter is correct, the etilqs_* files aren't part of those (and never looked at during startup/recovery). The WAL etc. are flushed to disk quite often, which obviously makes sense. I haven't checked whether the etilqs_* files are actually flushed to disk as well, or if some of that could be caught by the OS cache (which would make sense for those). Thanks for taking a look into this matter! It might not only make a difference for disk write activity, but possibly also for performance (CPU load). I'd really love to see this awesome product become even better!

I'm currently trying out Resilio Sync Pro and I'm considering to buy it, as it perfectly fits my needs except for one thing: There seems to be an excessive amount of seemingly useless disk write traffic to be going on in the background (on Windows 10), which amounts to tens of gigabytes a day and might shorten SSD lifespan significantly. I'm talking about roughly 10 million write operations per hour according to Sysinternals Process Monitor. I'm well aware that there are some things that just need to be written to disk, and that modern SSDs can take quite a lot of writes. I don't want to start a flame war on what SSDs can or cannot take here. What annoys me is just the nature of most of the data being written, which seems completely useless to me: Quite a lot of it is going to a HUGE sync.log file. That log seems extremely verbose and easily accounts for hundreds of megabytes per day. I can understand why that might help in diagnosing issues, but I think there should be a way to turn that down to a sane level (or completely off) somewhere in the power user settings if everything is just working fine. EDIT: Apparently this behavior can be configured using %APPDATA%\Resilio Sync\debug.txt which seems to be a bit mask of enabled debug sources / log levels. What seems even more useless is an incredible amount of writes to very small %TEMP%\etilqs_* files, which are quite likely SQLite temporary tables. Given their volatile nature and the fact that they're tiny (<10KB usually) it might adequate to just keep them in RAM. AFAIK SQLite has a configuriation option for that, but there's nothing that the end user can do about it. EDIT: More info: https://www.sqlite.org/tempfiles.html#the_sqlite_temp_store_compile_time_parameter_and_pragma Are there any plans to mitigate those issues? These really seem easy enough to fix to me, although I guess that quite a large part of the user base doesn't really bother about this as much. Apart from those nitpicks this really seems like the ideal product to fit my needs, which no other product on the market can right now. Realtime peer-to-peer sync with Selective Sync and the ability to handle large amounts of files without any apparent performance issues are the killer feature here.