You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Using GitPython==0.3.2.RC1 with Python 2.7.1 on a Fedora 15 with "/etc/security/limits.conf" nproc set to 25000 for the user running the test.
After repetitive calls to different repositories using git.GitCmdObjectDB, the system resources decrease reaching a "OSError: [Errno 24] Too many open files" exception.
This does not happen with the default odbt (gitdb).
Digging into the problem, lsof outputs lots of "pipe" processes. Possibly, the calls to git command using subprocess modure aren't properly deallocated.
Is this a bug?
Thanks,
The text was updated successfully, but these errors were encountered:
If the GitCmdObjectDB is used, each repository instance will have two slave git processes which stream objects from the git database. If for some reason the repositories don't get deallocated, the slave processes would stay intact. As a workaround, you might check when you create repositories, and if they get deleted at some point.
However, there might also be a bug of some kind, in theory at least the destruction of the repository should close the pipes which connect to the slave processes, which in turn should shut down the attached process. Something might not work there.
Using GitPython==0.3.2.RC1 with Python 2.7.1 on a Fedora 15 with "/etc/security/limits.conf" nproc set to 25000 for the user running the test.
After repetitive calls to different repositories using git.GitCmdObjectDB, the system resources decrease reaching a "OSError: [Errno 24] Too many open files" exception.
This does not happen with the default odbt (gitdb).
Digging into the problem, lsof outputs lots of "pipe" processes. Possibly, the calls to git command using subprocess modure aren't properly deallocated.
Is this a bug?
Thanks,
The text was updated successfully, but these errors were encountered: