19.11.2012 20:08, Robert Fox kirjoitti: > On Mon, 2012-11-19 at 16:57 +0200, Sander Lepik wrote: >> 18.11.2012 17:09, Thomas Spuhler kirjutas: >>> On Sunday, November 18, 2012 05:20:09 AM Anssi Hannula wrote: >>>> 18.11.2012 13:30, Thierry Vignaud kirjoitti: >>>>> On 18 November 2012 11:38, Anssi Hannula <[email protected]> wrote: >>>>>>> Bug has been filed: >>>>>>> https://bugs.mageia.org/buglist.cgi?quicksearch=looping+forever >>>>>>> >>>>>>> Updating Cauldron is broken as of this morning - >>>>>>> >>>>>>> [root@mainfox rfox]# urpmi --auto-update -v >>>>>> [...] >>>>>> >>>>>>> detecting looping forever while trying to resolve dependancies. >>>>>>> Aborting... Try again with '-vv --debug' options >>>>>>> at >>>>>>> /usr/lib/perl5/vendor_perl/5.16.2/x86_64-linux-thread-multi/URPM/Resol >>>>>>> ve.pm line 1061. unlocking rpm database >>>>>>> unlocking urpmi database >>>>>> It seems the loop threshold Resolve.pm:1061 is way too low and detects >>>>>> looping when there in fact isn't any. I bumped it up by several orders >>>>>> of magnitude locally and then urpmi worked again. >>>>> To what value did you bump it? >>>> To 100000, but that was just a random value since I mainly just wanted >>>> the upgrade to work. I don't know what would be a reasonable value. >>> Did you upgrade it on the server? if yes, this didn't solve the problem. >>> >> No he didn't. Thierry, are you sure that your limit is correct? If >> 100000 works then i don't see it as an endless loop. Current value seems >> to be too low and there are soon as many bugs about update problems than >> we have about broken draksnapshot. >> >> -- >> Sander >> > > I changed the value to 10000 as well and it worked. So it is definitely > too low of a limit (1000) -
As another random datapoint, I just tested (--test) upgrading one of my mga2 boxes to cauldron, and the count was 8213 there. The loop took approx. 3 seconds to execute, so we could probably make the value e.g. 50000 or 100000? -- Anssi Hannula
