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.
On Fri, 01 Sep 2006 12:57:16 -0400, Chuck Kirschman <Chuck.Kirschman at Nosp_am.bentley.com> wrote: >I think this is one of those things that is only seen by larger users, >but we definitely found it to be a problem. We regularly got up over >100 MB history files, and some repositories cleared 500 MB and were >heading for 1 GB. As near as I can tell, CVS still does not have a >solution to this problem. What we ended up doing was writing a simple >script to keep roughly the last 3 months of data; usually 30-50 MB. >This had substantial performance benefits when running a "cvs history" >command, but smaller effects on other operations. > >Rather than lose information, we stored the pieces we clipped out of the >history files in other files with a time stamp in the name. So now a >"cvs history" will only show data for the past 3 months. However, when >we need to go back further, we can use a combination of greps to extract >the information into a single history file that we can then query. > >I do not believe there is any way for the history file in one repository >to influence any other repository. > If you have a recent CVSNT server, then why not enable auditing and switch off the history file stuff altogether? With a proper database you get so much better possibilities to search for stuff and create all kinds of reports. HTH /Bo (Bo Berglund, developer in Sweden)