Re: Child pid '4787' is dead
Dne 6.9.2016 v 23:22 Miro Hrončok napsal(a): > My builds in [python26], IDs 450012 and 450019 end weird on F23, while > succeeding on other Fedoras. > > The last lines of build.log.gz are: > + exit 0 > Child pid '4787' is dead > Child dead, killing orphans > Child return code was: 0 > > While on other Fedoras: > + exit 0 > Child return code was: 0 > > The [koji] build is fine (at least for non-arm now). > > The RPM files are there in the output directory, but not installable from the > repo. > > Is that some problem in Copr? Not Copr problem. May be a mock problem. Lookin in mock code: i_rdy, o_rdy, e_rdy = select.select(fds, [], [], 1) if not i_rdy and not o_rdy and not e_rdy: if child and child.poll() is not None: logger.info("Child pid '%s' is dead", child.pid) To me is looks like a race condition: select() waited 1 second and returned empty list and exactly at the same time the process ended so pool() returned "not None". The strange is that it happened twice in row for you. -- Miroslav Suchy, RHCA Red Hat, Senior Software Engineer, #brno, #devexp, #fedora-buildsys ___ copr-devel mailing list copr-devel@lists.fedorahosted.org https://lists.fedorahosted.org/admin/lists/copr-devel@lists.fedorahosted.org
Re: Child pid '4787' is dead
On 8.9.2016 13:59, Miroslav Suchý wrote: Dne 6.9.2016 v 23:22 Miro Hrončok napsal(a): My builds in [python26], IDs 450012 and 450019 end weird on F23, while succeeding on other Fedoras. The last lines of build.log.gz are: + exit 0 Child pid '4787' is dead Child dead, killing orphans Child return code was: 0 While on other Fedoras: + exit 0 Child return code was: 0 The [koji] build is fine (at least for non-arm now). The RPM files are there in the output directory, but not installable from the repo. Is that some problem in Copr? Not Copr problem. May be a mock problem. Lookin in mock code: i_rdy, o_rdy, e_rdy = select.select(fds, [], [], 1) if not i_rdy and not o_rdy and not e_rdy: if child and child.poll() is not None: logger.info("Child pid '%s' is dead", child.pid) To me is looks like a race condition: select() waited 1 second and returned empty list and exactly at the same time the process ended so pool() returned "not None". The strange is that it happened twice in row for you. Well, three times in a row. Trying again. -- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok ___ copr-devel mailing list copr-devel@lists.fedorahosted.org https://lists.fedorahosted.org/admin/lists/copr-devel@lists.fedorahosted.org