On Wed, May 15, 2013 at 10:57 AM, Maciej (Matchek) Bliziński <[email protected]> wrote: > By the way, you might have not noticed that your first run actually > succeeded. You saw an error message, so you didn't check the $? > variable or the catalog content, but I bet you €100 it was correct. > > Why? Because of the new retry code. I noticed that on my VM, when > running a lot of HTTP queries for a long time, a query might > occasionally fail with no obvious reason. Since all RESTful calls are > idempotent, it's safe to retry, so I added some rudimentary retry > logic a few weeks ago. In your first run, one of the queries failed > with a HTTP 500, but it waited a few seconds, tried again, and the > retried run has succeeded, so the process could move forward. So this > failure was so transient that didn't even break your csw-upload-pkg > run! Yay REST + idempotence.
I did notice the text about retry but I wasn't sure about how successful that was so I re-ran the whole set. It looked to me like the retry also got a code 500 but maybe it was just that it was called with that error or something. >> INFO:root:Retry, exception: >> http://buildfarm.opencsw.org/releases/catalogs/unstable/i386/SunOS5.10/e5bcb76cbce9bbc7d9f8bcaad5485422/ >> - HTTP code: 500 _______________________________________________ maintainers mailing list [email protected] https://lists.opencsw.org/mailman/listinfo/maintainers .:: This mailing list's archive is public. ::.
