Update of sr #109215 (project administration):
Assigned to: None => rwp
Open/Closed: Open => Closed
_______________________________________________________
Follow-up Comment #1:
Thank you for your report. What you comment upon is likely true. That the
old vcs git server was overloaded at that time and then recovered later.
One detail is missing. Was this using the git:// service or was this over ssh?
I assume over git:// as that would match the typical symptom.
Note that contrary to my last note that said git had moved to the new server I
had to move it back because of a missing piece of functionality. --depth 1
for shallow checkouts did not work with the git server. I am working on it
and hope to have that resolved soon. Until then I moved things back to the
old server. It was on the old server when you noted the recent problem.
Unfortunately at this moment it just looks to have been general overloading of
the server with requests from the Internet more than it could handle. The old
server is using xinetd to spawn git-daemon and the xinetd drops connections
when they have reached their process number limits. The new server uses
git-daemon itself to manage incoming requests and will attempt to queue
incoming connections up through the OS network accept limits and generally
should behave better.
_______________________________________________________
Reply to this item at:
<http://savannah.gnu.org/support/?109215>
_______________________________________________
Message sent via/by Savannah
http://savannah.gnu.org/
