[MacPorts] #52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute
#52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute ----------------------------+-------------------------- Reporter: ryandesign@… | Owner: ryandesign@… Type: defect | Status: new Priority: Normal | Milestone: Component: server/hosting | Version: Keywords: | Port: ----------------------------+-------------------------- https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/7064 I see in the log: {{{ Warning: xcodebuild exists but failed to execute }}} and {{{ Warning: Xcode does not appear to be installed; most ports will likely fail to build. }}} Which causes "System cc" to be the compiler that's used: {{{ DEBUG: Using compiler 'System cc' }}} Which causes at-spi2-core to fail to build with the universal variant: {{{ lipo: /opt/local/var/macports/build /_opt_bblocal_var_buildworker_ports_build_ports_gnome_at-spi2-core/at- spi2-core/work/.tmp/cc3UStyY.out and /opt/local/var/macports/build /_opt_bblocal_var_buildworker_ports_build_ports_gnome_at-spi2-core/at- spi2-core/work/.tmp/ccyIbdDh.out have the same architectures (x86_64) and can't be in the same fat output file }}} Which causes wine to be unable to be installed. I am seeing this only on snowleopard-x86_64-legacy, not on snowleopard-i386-legacy nor snowleopard-x86_64 nor the other workers. It was my intention to set up all the buildbot workers the same way. This problem appears to have started in [https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/6874 build 6874]. [https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/6873 Build 6873] says: {{{ DEBUG: Using compiler 'Xcode GCC 4.2' }}} but as of [https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/6874 build 6874] we start to see: {{{ DEBUG: Using compiler 'System cc' }}} I don't know what caused this. We ran into the same problem on the previous Snow Leopard buildbot worker a couple years ago; see #46099. I couldn't figure out what caused that either; it solved itself. -- Ticket URL: <https://trac.macports.org/ticket/52543> MacPorts <https://www.macports.org/> Ports system for the Mac operating system
#52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute -----------------------------+-------------------------- Reporter: ryandesign@… | Owner: ryandesign@… Type: defect | Status: closed Priority: Normal | Milestone: Component: server/hosting | Version: Resolution: fixed | Keywords: Port: | -----------------------------+-------------------------- Changes (by ryandesign@…): * status: new => closed * resolution: => fixed Comment: Not knowing what else to try, I've rebooted this worker. This appears to have helped: [https://build.macports.org/builders/ports-10 .6_x86_64_legacy-builder/builds/7092 build 7092] shows: {{{ DEBUG: Using compiler 'Xcode GCC 4.2' }}} Short of committing a change to the at-spi2-core portfile, I don't know how to get the buildbot to clear the failed build from the failcache. Just forcing a build of at-spi2-core on the buildbot web site will presumably only clear the non-universal build from the failcache, and forcing a build of wine again will not attempt to build at-spi2-core again because it's in the failcache. -- Ticket URL: <https://trac.macports.org/ticket/52543#comment:2> MacPorts <https://www.macports.org/> Ports system for the Mac operating system
#52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute -----------------------------+-------------------------- Reporter: ryandesign@… | Owner: ryandesign@… Type: defect | Status: closed Priority: Normal | Milestone: Component: server/hosting | Version: Resolution: fixed | Keywords: Port: | -----------------------------+-------------------------- Changes (by devans@…): * cc: cal@… (added) Comment: Replying to [comment:2 ryandesign@…]:
Short of committing a change to the at-spi2-core portfile, I don't know how to get the buildbot to clear the failed build from the failcache. Just forcing a build of at-spi2-core on the buildbot web site will presumably only clear the non-universal build from the failcache, and forcing a build of wine again will not attempt to build at-spi2-core again because it's in the failcache. If I'm looking at the relevant code, it appears that, if you could generate/determine the failcache hash for the given port and variant, you could manually remove the corresponding entry from the failcache directory. This is essentially what [http://svn.macports.org/repository/macports/contrib/mp-buildbot/functions failcache_success() in contrib/mp-buildbot/functions] does. Maybe it would be useful for admin purposes to have a tool that would do this or, even more elegantly, add something to an appropriate buildbot display that would allow someone with login access to do this for a given failed build.
-- Ticket URL: <https://trac.macports.org/ticket/52543#comment:3> MacPorts <https://www.macports.org/> Ports system for the Mac operating system
#52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute -----------------------------+-------------------------- Reporter: ryandesign@… | Owner: ryandesign@… Type: defect | Status: closed Priority: Normal | Milestone: Component: server/hosting | Version: Resolution: fixed | Keywords: Port: | -----------------------------+-------------------------- Comment (by ryandesign@…): Replying to [comment:3 devans@…]:
If I'm looking at the relevant code, it appears that, if you could generate/determine the failcache hash for the given port and variant, you could manually remove the corresponding entry from the failcache directory. This is essentially what [http://svn.macports.org/repository/macports/contrib/mp-buildbot/functions failcache_success() in contrib/mp-buildbot/functions] does.
True. From [https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/7085 build 7085]: {{{ port at-spi2-core +universal 942bd1ec04930577a3b0c8857ada6c840e6f793a2a65f0a52bfcb03b624f6c7b previously failed in build https://build.macports.org/builders/ports-10 .6_x86_64_legacy-builder/builds/7064 }}} I removed that file from the failcache and started [https://build.macports.org/builders/ports-10.6_x86_64_legacy- builder/builds/7107 build 7107].
Maybe it would be useful for admin purposes to have a tool that would do this or, even more elegantly, add something to an appropriate buildbot display that would allow someone with login access to do this for a given failed build.
I think the intention is that you can force a new build of the port, which ignores the failcache. The problem is that here the failed build is universal, and we don't have a way to specify variants when forcing a build. Maybe we should add that, though I'm wary of letting people build arbitrary variants on the buildbot. Maybe we can add checkboxes for the two specific variants that we're considering making the buildbot build automatically: universal and quartz. Though I'm not sure how much we can customize the web pages. -- Ticket URL: <https://trac.macports.org/ticket/52543#comment:4> MacPorts <https://www.macports.org/> Ports system for the Mac operating system
#52543: snowleopard-x86_64-legacy buildbot worker: xcodebuild exists but failed to execute -----------------------------+-------------------------- Reporter: ryandesign@… | Owner: ryandesign@… Type: defect | Status: closed Priority: Normal | Milestone: Component: server/hosting | Version: Resolution: fixed | Keywords: Port: | -----------------------------+-------------------------- Comment (by cal@…): The scripts already support building arbitrary variants, we just haven't made this option available to developers in the buildbot web interface. We already trust our devs to not build weird things on the buildbot, I don't see how the variants are different. If somebody abuses building of variants, we can still shout at them. -- Ticket URL: <https://trac.macports.org/ticket/52543#comment:5> MacPorts <https://www.macports.org/> Ports system for the Mac operating system
participants (1)
-
MacPorts