On 1/18/08, James Antill <[EMAIL PROTECTED]> wrote: > > On Fri, 2008-01-18 at 02:49 -0500, Jay Soffian wrote: > > This patch does two things: > > > > 1) It fixes retries. YumOuput.failureReport() re-raising the > > errobj.exception prevents urlgrabber from retrying. > > This is for the non-mirror failure mode, if you keep retrying here you > might be re-downloading something that is "working" from the HTTP POV, > but failing the checks post download. > So you'd need some way to distinguish between a temporary network > problem and a data problem.
I'm not sure what you mean by "non-mirror failure mode". failureReport() is registered as the URLGrabber failure_callback, so it's called for any download failures, regardless of the number of URLs configured for the repo. > > 2) In order to be friendly to the HTTP server, it adds a random > > exponential backoff delay between retry attempts. > > This seems like a bad idea, much better to just go for between 2 and 16 > seconds or something. Uh, it's worked well enough in Ethernet for 30+ years. It seems to work well in my experience as well (I've stress tested against a local server). > > This is against 3.0.x. > > That's _old_, there's going to be a lot of stuff fixed by just > upgrading. According http://linux.duke.edu/projects/yum/download.ptml there are several stable releases: - 3.2.x - 3.0.x - 2.6.x - 2.4.x Are you not interested in fixing bugs in older releases? At least the 3.0.x branch is in use by EL5. j. _______________________________________________ Yum-devel mailing list [email protected] https://lists.dulug.duke.edu/mailman/listinfo/yum-devel
