On our setup, if any more than about 20 people try to svn update simultaneously, the whole thing grinds to a halt.
We can limit our number of connections to 20, and if you try to svn update and there are already 20 connections, you are told to try again later, but I was hoping开发者_C百科 to set up a more elegant user experience.
Before I attempt to code a bespoke solution, is there anything off-the-shelf that will put additional svn update requests into a queue, that will begin your svn update as soon as the number of connections drops below 20 (and you are first in the queue) and that will show you where you are in the queue?
Thanks
Tim
the changes to the network transport used in SVN sounds like what you want. Currently SVN uses the neon http library and all requests for data are bundled up into a single REQUEST. The devs have been working on adding SERF support to svn too (well, its always been there but not used very much) and this will fetch data using a lot of GET or POSTs. This should be much more scalable (though potentially less performant overall if you are fetching a very large number of very small files).
So, you can try to switch to using the serf library by setting the http-library variable in your server configuration. Its not as reliable as neon in the current versions of SVN, but I think it'll be the default library for v1.7
精彩评论