Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
Hello all, here I am again with the questions we've got having moved from cvs to cvsnt. In the repository we have some "service" files which are used in our build system. It's turned out that sometimes cvs co -p -r tag module/proj/file fails claiming that no such tag exists. At the same time the following sequence works fine: cvs co module/proj/file cd module/proj cvs up -r tag Then I wrote simple script which tries to get file by each of existing tags and it has failed on 48 tags of 81. Again, 'cvs up -r' was just fine for all the tags. But 'cvs up -r xxx module/proj' brought something new - checkout command executed extremely slow. Actually, this script for 81 tags took several hours to complete (though 'proj' contains only two files by several bytes). Weird that cvsnt process on the server side this time was actively working with files in other directories of the repository (not only in current module, but in the second module of the repository too). And even some more weird that such a behaviour disappeared by its own. I just noticed that after script completed, everything became ok - tags were on their place, and project was no so huge as it pretended. Other people from our staff already reported (very occasionly) "slow-to-death checkouts" after moving to cvsnt, hence it doesn't look like a "moon phase" or something. Have any ideas where to start from? Which server-side trigger scripts may (in theory) produce such a behaviour? What additional information should I supply? Client: CVSNT 2.0.41a Win32 Server: CVSNT 2.0.41a Linux -- Best regards, Iakov mailto:iglubokiy at infotecs.ru