Assumption: (There are only two parts (-n, -l) of "repo sync" requiring
multiple CPU cores to use a distributed server system effectively)
repo sync = [repo sync -n] + [repo sync -l]
repo sync -n = “network only”. (Downloading git DB ".git"s)
repo sync -l = “local only”. (Checking out source codes - actual file
writing)
Please, correct me if I am wrong in some point.
Feature request:
Requesting a feature to enable the parallel CPU process for checking out source
codes (repo sync -l). If there is another option to enable the parallel CPU
process for checking out source codes, we will be able to use our distributed
server environment more effectively.
If there are other options or ways to support this feature,
Reason for requesting this feature:
repo sync -j option uses multiple CPU cores, but only during "repo sync -n"
part. When it processes "repo sync -l" part, it will always use a single core.
Even if we use a wide distributed server environment, there are some
limitations to use full CPU cores during "repo sync". With this limitation,
all we can do is to extend client servers to support more users.
What steps will reproduce the problem?
1.repo init -u https://android.googlesource.com/platform/manifest
2.repo sync
3.See how many CPU cores are being used for each steps ([repo sync -n] + [repo
sync -l]).
Original issue reported on code.google.com by eric....@lge.com on 11 Nov 2014 at 5:47
Original issue reported on code.google.com by
eric....@lge.com
on 11 Nov 2014 at 5:47