CVS Outage

Zing zing at fastmail.fm
Tue Apr 22 01:52:50 UTC 2008


On Mon, 21 Apr 2008 09:56:05 -0600, Stephen John Smoogen wrote:

> On Mon, Apr 21, 2008 at 8:56 AM, Gary Thomas <gary at mlbassoc.com> wrote:
>>
>> Jesse Keating wrote:
>>
>> > On Sat, 2008-04-19 at 15:05 -0400, Jesse Keating wrote:
>> >
>> > > There will be an outage starting at 2008-04-19 18:00 UTC.  Initial
>> > > estimates show that 30 hours will be needed to complete the
>> > > branching, however this may drastically be reduced during the full
>> > > run.
>> > >
>> > >
>> > Unfortunately this outage is still ongoing.  I failed to get a
>> > notification when the first leg finished around 2am my time and lost
>> > 6~ hours to sleep.  All the PackageDB settings are done (barring a
>> > few failures) and we're well into the packages that start with "p"
>> > for making the branches within CVS.  I expect it to take another half
>> > a day or so to finish.
>> >
>> > Future Recommendation:
>> >  Enable methods of doing mass branching without incurring an outage.
>> > Make pkgdb be able to bypass mail sending, and make CVS be able to
>> > bypass mail sending for these operations.  Will take longer overall,
>> > but no outage.
>> >
>> >
>>  Recommendation - switch to SVN (quite painless).  All of these things
>>  are pretty simple to implement.  Things like tagging become O(0) [i.e.
>>  constant and *fast*]
>>
>>
> All Version Control Systems are broke when they deal with large
> projects... even SVN has its limitations.
> 
> The big issue is that the religious wars that get involved with any of
> the systems. People tantrum if their system is not chosen and every time
> a problem comes up with the new system they will point out why choosing
> XCS was boneheaded. This bogs down the cycle more than anything else.

Cleary someone from the fedoraproject needs to design a new vcs tailored 
just for fedora.  <ducks>





More information about the fedora-devel-list mailing list