martijnverburg at gmail.com
Mon Oct 29 19:34:51 UTC 2018
FWIW - we’ve seen similar download speeds from our git clones of the
mercurial forests (hosted at GitHub) and interestingly it’s decent speeds
for our binary releases as well. GitHub’s CDN capabilities seem pretty
good, although when they have an outage it certainly hurts.
The clone is a bit slower for full 8 update forest clone (loads more
history), but it’s still good overall.
@Joe - I’ll echo Patrick’s offer to help. Whatever we can do / experiment
on your behalf let us know, especially if you’re looking to migrate the 8u
repo which has a number of git/hg challenges with regards to history etc
that we are working around, it’s been interesting...
On Mon, 29 Oct 2018 at 16:51, joe darcy <joe.darcy at oracle.com> wrote:
> Hi Patrick,
> The times you're seeing are consistent with the measurements we've made.
> Cloning from hg.openjdk can take upwards of 30 minutes in Europe and
> other locations far (in internet terms) from the server. Clones from
> github tend to be 1 to 2 minutes at all the locations we've measured,
> given reasonable equipment.
> On 10/27/2018 10:54 AM, Patrick Reinhart wrote:
> > Hi Joe,
> > Wow !!
> > [pr at wsccuw01 openjdk]$ time git clone https://github.com/openjdk/jdk
> > [...]
> > real 0m49.661s
> > user 0m40.437s
> > sys 0m6.495s
> > versus:
> > [pr at wsccuw01 openjdk]$ time hg clone https://hg.openjdk.java.net/jdk/jdk
> > [...]
> > real 35m0.594s
> > user 5m48.109s
> > sys 0m33.165s
> > -Patrick
> > Am 25.10.18 um 22:54 schrieb joe darcy:
> >> On 10/25/2018 1:29 PM, Patrick Reinhart wrote:
> >>> Hi Joe,
> >>> I will get my a clone and see how fast that goes compared to the
> >>> mercurial way...
> >>> .. and hey: I was ranked as #33 contributor :-)
> >>> -Patrick
> >> Thanks Patrick.
> >> For all the download testing we've done, we've seen healthy speedups
> >> compared to hg :-)
> >> -Joe
Cheers, Martijn (Sent from Gmail Mobile)
More information about the skara-dev