On 2025-04-28 18:11, Andreas Stieger wrote:
> However, I am not aware of anything breaking for the server itself, access to the on-disk repository date is purely path based unless I am mistaken.
We noticed the issue that all had the same UUID while we were trying to figure out a problem we encountered with our set up.
We automate a lot of edits and have created our own SVN client program that communicates with the servers. One functionality is that upon creation of a new repository through a REST service (which internally then copies the template repository as described) the client also SVN imports a file to a related repository (e.g. create 'AB-123' and 'AB' gets updated to reflect that '123' exists).
We tested that newly developed client feature with the same repository names multiple times, deleting the test repositories and recreating them to test the feature again. And what happened was that according to the TortoiseSVN repo browser, no file was imported, according to the SVN log, it was imported but into the wrong directory, you could not open the SVN log inside the repo browser because it would throw errors, etc. Opening the TortoiseSVN repo browser from the server's file system however, it showed correct commits and contents.
Therefore we suspect it was some kind of caching issue in the SVN server (WANdisco) due to same path same UUID same commit number; after restarting the SVN server, the issue went away. But I don't know if WANdisco is a "typical server implementation", or if the issue was something deeper, or if the issues are related at all or pure coincidence, etc.
Thanks a lot for the responses so far, Daniel and Andreas. We will evaluate whether we want to do something for the existing repositories.
Christoph
