> Apparently another abf issue, as it is not the first time it says "metadata already queued for regeneration" but looks like only regenerates it after someone from abf interfering, i.e. after like 8+ hours.
See: https://abf.io/platforms/cooker/repositories/main/edit Current status | Last status of regeneration metadata | Last log of regeneration metadata | Last regeneration of metadata No actions| Completed succesfully | regeneration.log<http://file-store.rosalinux.ru/api/v1/file_stores/841163649795853f2f5e0b7ae2524d2ba78b8815.log?show=true> | 2014-04-05 19:23:37 UTC Logs: http://file-store.rosalinux.ru/api/v1/file_stores/841163649795853f2f5e0b7ae2524d2ba78b8815.log?show=true Regeneration works. Please, be patient. Let's go discuss "Do not remove packages required by others in publish and implement garbage-collector for orphan rpms" in https://abf.io/abf/abf-ideas/issues/128 Thanks, Alexey On Sat, Apr 5, 2014 at 10:38 PM, Александр Бурмашев < [email protected]> wrote: > The problem will never happen anymore if libs + all packages, requiring > them, are built before publishing the updated library into the repository. > > > > > 2014-04-05 22:23 GMT+04:00 Paulo César Pereira de Andrade < > [email protected]>: > > 2014-04-05 15:15 GMT-03:00 Bernhard Rosenkränzer <[email protected]>: >> > Hi, >> > >> > >> > On 2014-04-05 18:02, Paulo César Pereira de Andrade wrote: >> >> >> >> I cannot work on cooker for 24 hours now, I pinged in irc when >> >> detected the problem but nobody commented :-( >> > >> > >> > That can happen at times -- for example, I was out in nowhere with >> Chwido >> > for the day (seems he likes the same remote places I do). >> > >> > >> >> The issue is that ncurses was updated, and it changed major of >> >> libncursesw. It did properly follow Mandriva libname specs changing >> >> the major of the package, e.g. changed libncursesw5 to libncursesw6, >> >> *but* when publishing, "abf" just removed all packages generated >> >> by the previous ncurses, that included libncursesw5. >> > >> > >> > That's the right thing to do, it's just wrong to have a binary package >> that >> > doesn't have a corresponding source package in the tree. >> > It does mean, however, that we have to be extra careful -- build the >> compat >> > package providing the old soname BEFORE building the new package with >> the >> > updated soname. >> >> Anything that can fail will fail, so this problem will repeat again and >> again >> if nothing is changed, but most times it does not break the chroot... >> >> > Maybe abf's publishing tests should be extended to catch this? Right >> now it >> > warns if a package can't be installed because it conflicts with itself >> or >> > the likes. A check for removing a .so.X that is still needed by >> something >> > else would be great (and an urpmq --whatrequires call shouldn't be too >> > expensive). >> >> I believe my suggestion idea is better described at >> https://abf.io/abf/abf-ideas/issues/128#comment4931 >> and the garbage collector (continuing 4931) at >> https://abf.io/abf/abf-ideas/issues/128#comment4932 >> >> >> Anyway, to fix the cooker issue, I hereby request some kind of >> >> abf access, maybe a way to upload files, e.g. rebuild bash >> >> locally with the newer libncursesw (but I would do it only for >> >> i586 and x86_64) and upload it, at least this way I would not >> >> be blocked from working for at least 24 hours. >> > >> > >> > Certainly not opposed to giving you more access (but I don't have the >> > privileges to do so myself) -- but you could have fixed this even >> without >> > more access, simply locate an older ncurses and re-publish it. That >> would >> > have killed libncursesw6, but nothing uses that yet. Would have been >> good >> > enough to get things running again so a compat package could be built >> (and >> > then ncurses 6 could be published again). >> >> Well, mdawkins did copy rpms from openmandriva2014.0 to cooker, but >> now metadata regeneration does not happen. Apparently another abf issue, >> as it is not the first time it says "metadata already queued for >> regeneration" >> but looks like only regenerates it after someone from abf interfering, >> i.e. >> after like 8+ hours. >> >> > ttyl >> > bero >> >> Thanks, >> Paulo >> >> > -- _________________________________ Best wishes, Your Vokhmin Alexey. e-mail: [email protected] skype: avokhmin
