On Thu, Feb 11, 2016 at 7:42 PM, Kay Schenk <kay.sch...@gmail.com> wrote: > > > On 02/11/2016 09:30 AM, Damjan Jovanovic wrote: >> On Wed, Feb 10, 2016 at 7:49 PM, Kay Schenk <kay.sch...@gmail.com> wrote: >>> >>> On 02/10/2016 09:41 AM, Damjan Jovanovic wrote: >>>> >>>> All the *nix buildbots are now green and should be building 100% reliably. >>> >>> YAY! >>> >> >> I jinxed it, many bots failed in ./bootstrap as they couldn't download >> dependencies. We should probably cache dependencies instead (ie. "cp >> external_deps/* build/ext_sources" before bootstrap and "cp >> build/ext_sources/* external_deps" after successful bootstrap). I'll >> try this on openoffice-linux64-nightly. > > oh well... yeah, I saw this. > > Yesterday, when you said you set up the builbots to fail on > bootstrap, I was going to say that even in my local build, the first > attempt at a bootstrap source sometimes fails but seems to right > itself the second time around usually from our URL2 sites. This > said, it could be that some of our buildbots, due to time out > issues, will cause failure with this > step. We might think of putting things back the way they originally > were for the ./bootstrap step but increasing the time. On my local > builds, of course, I just sit and wait. >
What I did didn't reduce ./bootstrap's resilience. It tries to download everything, and only when it's gone through all files, exits with an error if something failed to download from all its URLs. The buildbots still failed despite ./bootstrap twice. Seems like a serious networking problem. I am trying even more caching. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
