[Maria-developers] Patches in 10.0 missing in 10.0-base, causing buildbot failures
Hi, I was looking at 10.0-base in Buildbot, in particular at the "update2" failures in all the .deb builds. I fixed one thing in Buildbot, but it still fails. I suspect it is because of these two (at least) commits: sergii@pisem.net-20121027121326-q87f5lt6ldxr1em6 sergii@pisem.net-20121020163522-uw2apo36c6nn6yi8 These were pushed to 10.0, while it seems they should instead have been pushed to 10.0-base (where they are missing). There are probably also other commits with same problem, eg. this one which is a 5.5 merge according to commit message: sergii@pisem.net-20121103112851-bvy1mlhirxbkedvf Any suggestions what to do to fix this and get 10.0-base working in Buildbot? - Kristian.
Hi, Kristian! On Dec 17, Kristian Nielsen wrote:
I was looking at 10.0-base in Buildbot, in particular at the "update2" failures in all the .deb builds.
I fixed one thing in Buildbot, but it still fails. I suspect it is because of these two (at least) commits:
sergii@pisem.net-20121027121326-q87f5lt6ldxr1em6 sergii@pisem.net-20121020163522-uw2apo36c6nn6yi8
These were pushed to 10.0, while it seems they should instead have been pushed to 10.0-base (where they are missing).
I can backport them to 10.0-base. But it's not particularly urgent, we aren't going to build packages from 10.0-base, right? I agree it should be done eventually, though Regards, Sergei
Sergei Golubchik <serg@askmonty.org> writes:
I can backport them to 10.0-base. But it's not particularly urgent, we aren't going to build packages from 10.0-base, right?
No, not urgent, agree. I've changes Buildbot so that the upgrade2 test is skipped in 10.0-base. So we get rid of the annoying failures. As you say, we're not going to release packages from 10.0-base anyway. With regard to the 5.5 merge to 10.0 rather than 10.0-base, seems igor already merged 5.5 to 10.0-base, and bzr --weave seems to mostly deal ok with that. Thanks, - Kristian.
Hello all just joined the development list recently, looking at doing a little bit of development maybe. Apologies in advance if this is the wrong list to ask my question. I have been working through the information at https://kb.askmonty.org/en/Getting_the_MariaDB_Source_Code/ and related pages. Being used to SVN and git, I am still finding my way around bzr, and finding it incredibly slow executing `bzr branch lp:maria`. I am not sure whether it is just me, but I am finding the kB/s is a long way below my connection speed. Anyway I decided to try and shortcut the process and download the branch starting tarball instead, except that all the links on the page http://askmonty.org/wiki/MariaDB:Download:Shared_Repo seem to be broken? Is the tar.gz shared repo information still valid? thanks, Andrew -- https://launchpad.net/~andymc73 http://blog.oldcomputerjunk.net Twitter: @andymc73 GPG: http://www.andrewmcdonnell.net/gpg.html
Andrew McDonnell <bugs@andrewmcdonnell.net> writes:
Being used to SVN and git, I am still finding my way around bzr, and finding it incredibly slow executing `bzr branch lp:maria`. I am not sure whether it is just me, but I am finding the kB/s is a long way below my connection speed.
It is not just you. Bzr is incredibly slow and CPU/memory hungry :-( It used to be far worse, I have seen initial checkout taking 40 hours (yes, *fourty* *hours*) of CPU time and 2GB ram to complete. Now I think it is down to one or two hours or so on a fast machine, though memory usage is probably still high. We're mostly stuck with bzr for now for legacy reasons. Fortunately, speed is tolerable with a shared repo once initial checkout is complete.
starting tarball instead, except that all the links on the page
http://askmonty.org/wiki/MariaDB:Download:Shared_Repo
seem to be broken? Is the tar.gz shared repo information still valid?
I've Cc'ed Daniel, who may be able to fix that. However, it's probably easiest to just create your own empty shared repo, and do the initial checkout overnight or something. Thanks for the notice, and good luck! - Kristian.
Kristian Nielsen <knielsen@knielsen-hq.org> writes:
I've Cc'ed Daniel, who may be able to fix that. However, it's probably easiest
BTW, it seems it is just the redirect that is broken, the base link seems to still work: http://ftp-stud.hs-esslingen.de/pub/Mirrors/mariadb/mariadb-shared-repo.tgz (though I'd still recommend doing your own initial checkout if you can live with it, to get up-to-date data). - Kristian.
participants (3)
-
Andrew McDonnell -
Kristian Nielsen -
Sergei Golubchik