On Thu, Sep 01, 2016 at 09:52:28AM +0100, Sascha Steinbiss wrote: > Hi Santiago, > > thanks for this useful info. > > > Status: successful genometools_1.5.9+ds-2_amd64-20160811T2351Z > > Status: successful genometools_1.5.9+ds-2_amd64-20160816T1151Z > > Status: successful genometools_1.5.9+ds-2_amd64-20160821T235102Z > > Status: failed genometools_1.5.9+ds-2_amd64-20160831T144105Z > > Status: failed genometools_1.5.9+ds-2_amd64-20160831T152720Z > > > > All those builds were done in machines with 768 MB of RAM > > (and 1GB of swap). > > > > Why it built ok in the past and now it does not build? > > Good question. It could have to do with overall memory availability on > this machine?
That "has" to be the reason, because I don't have a better theory. But the only task those machines have is building packages. The timestamp for /etc/fstab is 2016-08-10, so they have used the same swap file (and therefore the same amount of swap) for all the builds. It is still unknown why it built in the past and not anymore. > Did the run you started on the machine with more memory > eventualy finish? Yes, on three other machines having 4 GB of swap and at least 3GB of RAM the build succeeded. > When I have some time I will try to reduce the amount of memory in my > building VM to find out what the critical value is. Thanks.
