extras-buildsys ChangeLog,1.196,1.197
Daniel Williams (dcbw)
fedora-extras-commits at redhat.com
Sun May 14 05:43:08 UTC 2006
Author: dcbw
Update of /cvs/fedora/extras-buildsys
In directory cvs-int.fedora.redhat.com:/tmp/cvs-serv2964
Modified Files:
ChangeLog
Log Message:
2006-05-14 Dan Williams <dcbw at redhat.com>
* Rework archjob handling. They are now processed from the owning
PackageJob object, and only one is spawned for the lifetime of the
PackageJob for each architecture (previously one was spawned for each
individual build job on the builder). Archjob UIDs are generated on
the server now rather than the builder.
* Correctly handle builders going away before they've had a chance to
notify the server that they have started an archjob. Previously, these
jobs would just be lost. This is the real reason for the rework of the
archjob handling above.
* On the builder, don't use die() to end jobs that have failed; do it
properly and upload files to the server if we weren't killed
* Deal with active builders whose hostnames can't be resolved when
the server starts
* Kill any existing jobs on builders when the server starts up
* Consolidate and simplify logging functions in PackageJob
* More pylint-induced cleanups
Index: ChangeLog
===================================================================
RCS file: /cvs/fedora/extras-buildsys/ChangeLog,v
retrieving revision 1.196
retrieving revision 1.197
diff -u -r1.196 -r1.197
--- ChangeLog 12 May 2006 04:10:45 -0000 1.196
+++ ChangeLog 14 May 2006 05:43:05 -0000 1.197
@@ -1,3 +1,28 @@
+2006-05-14 Dan Williams <dcbw at redhat.com>
+
+ * Rework archjob handling. They are now processed from the owning
+ PackageJob object, and only one is spawned for the lifetime of the
+ PackageJob for each architecture (previously one was spawned for each
+ individual build job on the builder). Archjob UIDs are generated on
+ the server now rather than the builder.
+
+ * Correctly handle builders going away before they've had a chance to
+ notify the server that they have started an archjob. Previously, these
+ jobs would just be lost. This is the real reason for the rework of the
+ archjob handling above.
+
+ * On the builder, don't use die() to end jobs that have failed; do it
+ properly and upload files to the server if we weren't killed
+
+ * Deal with active builders whose hostnames can't be resolved when
+ the server starts
+
+ * Kill any existing jobs on builders when the server starts up
+
+ * Consolidate and simplify logging functions in PackageJob
+
+ * More pylint-induced cleanups
+
2006-05-12 Dan Williams <dcbw at redhat.com>
* builder/Builder.py
More information about the fedora-extras-commits
mailing list