Merge lp:~bcsaller/pyjuju/local-troubleshooting into lp:pyjuju

[Benjamin Saller]

local provider troubleshooting aid

Includes docs and a small script to collect information about the local machines lxc environment for bug reports.

[lp:~bcsaller/pyjuju/local-troubleshooting]

406. By Benjamin Saller on 2011-10-13

note about container name on troubleshooting script

407. By Benjamin Saller on 2011-10-14

fixed some typos in the script

[William Reade (fwereade)]

[0]

+When providing bug reports or including issues the output of the both
+the master-customize.log mentioned above and the result of the
+devel-tools/juju-inspect-local-provider script.

Parse error. Perhaps "When reporting bugs or issues, please include both the master-customize.log (described above) and the output of the devel-tools/juju-inspect-local-provider script."?

Otherwise LGTM

review: Approve

[Kapil Thangavelu (hazmat)]

[0]
37 +When you first attempt to deploy a unit the time consuming process of
38 +creating a unit will begin

Services are deployed, service units are added/created

[1]

144 +virsh net-list --all

This should probably be delocalized LC_ALL="C"

review: Approve

=== modified file 'docs/source/provider-configuration-local.rst'
--- docs/source/provider-configuration-local.rst	2011-10-07 20:04:21 +0000
+++ docs/source/provider-configuration-local.rst	2011-10-14 23:44:24 +0000
@@ -41,11 +41,51 @@
 Provider specific options
 =========================
 
-  data-dir: 
-    Directory for zookeeper state and log files. 
-   
-
-
- 
-
-
+  data-dir:
+    Directory for zookeeper state and log files.
+
+
+
+Troubleshooting
+===============
+
+Local development is still a young and evolving feature and because of
+the many complex system interactions its possible you'll encounter
+some speed bumps along your path. There are a number of log files you
+should be aware of when trying to understand how the system behaves.
+
+Once``juju bootstrap`` has run you'll have a directory in place on
+your filesystem at the location specified in your
+environments.yaml. In here you'll find a number of files related to
+the runtime of the local deployment system. The first step is to
+verify that the machine-agent.log includes a line similar to::
+
+    2011-10-05 14:43:56,327: juju.agents.machine@INFO: Machine agent started id:0 deploy:<class 'juju.machine.unit.UnitContainerDeployment'> provider:'local'
+
+When you first attempt to deploy a unit the time consuming process of
+creating a unit will begin. The first time you run this it can taGke
+quite some time and if your impatient and need to check on the process
+`lxc-ls` will show you which templates are created or being created
+and::
+
+    pgrep lxc| head -1| xargs watch pstree -alU
+
+will give you a refeshing view of the first container as its built.
+
+Once a container is built you should have access to it via ssh through
+the `juju ssh` as normal.
+
+If services are not running properly as exposed by `juju status` the
+log file in *data-dir/units/master-customize.log* should provide insight
+into the nature of the error.
+
+When providing bug reports or including issues the output of the both
+the master-customize.log mentioned above and the result of the
+devel-tools/juju-inspect-local-provider script. If you are not running
+a development checkout this script should be located at
+*/usr/lib/juju/juju/misc/devel-tools/juju-inspect-local-provider*
+
+By passing the name of the container as an argument to that script
+additional information abou the container will be included in the
+output as well. To find the name of the container in question `lxc-ls`
+can be used.
 
=== modified file 'juju/lib/lxc/data/juju-create'
--- juju/lib/lxc/data/juju-create	2011-10-12 18:08:30 +0000
+++ juju/lib/lxc/data/juju-create	2011-10-14 23:44:24 +0000
@@ -146,3 +146,5 @@
 setup_juju
 # setup_juju ensures sudo is installed which is needed for setup_users
 setup_users
+
+echo "Container Customization Complete"
 \ No newline at end of file
 
=== modified file 'juju/providers/local/__init__.py'
--- juju/providers/local/__init__.py	2011-10-12 18:08:30 +0000
+++ juju/providers/local/__init__.py	2011-10-14 23:44:24 +0000
@@ -70,6 +70,13 @@
             raise ProviderError("Missing packages %s" % (
                 ", ".join(sorted(list(missing)))))
 
+        # Store user credentials from the running user
+        try:
+            public_key = get_user_authorized_keys(self.config)
+            public_key = public_key.strip()
+        except LookupError, e:
+            raise ProviderError(str(e))
+
         # Get/create directory for zookeeper and files
         zookeeper_dir = os.path.join(self._directory, "zookeeper")
         if not os.path.exists(zookeeper_dir):
@@ -122,13 +129,6 @@
         hierarchy = StateHierarchy(client, admin_identity, "local", "local")
         yield hierarchy.initialize()
 
-        # Store user credentials from the running user
-        try:
-            public_key = get_user_authorized_keys(self.config)
-            public_key = public_key.strip()
-        except LookupError, e:
-            raise ProviderError(str(e))
-
         # Startup the machine agent
         pid_file = os.path.join(self._directory, "machine-agent.pid")
         log_file = os.path.join(self._directory, "machine-agent.log")
 
=== added directory 'misc/devel-tools'
=== added file 'misc/devel-tools/juju-inspect-local-provider'
--- misc/devel-tools/juju-inspect-local-provider	1970-01-01 00:00:00 +0000
+++ misc/devel-tools/juju-inspect-local-provider	2011-10-14 23:44:24 +0000
@@ -0,0 +1,63 @@
+#!/bin/bash -x
+
+# Gather a collection of data into an output file to make
+# debugging any possible issues with the local provider simpler
+
+if [ ! `id -u` == '0' ]; then
+	echo "This script should be run as root"
+	exit 1;
+fi
+
+if [ ${#} = 1 ]; then
+    image_name=$1
+fi
+
+# 11.10 (Oneiric) is the first supported release for the local provider due
+# to its improved LXC support
+source /etc/lsb-release
+major_release=`echo $DISTRIB_RELEASE | cut -d . -f 1`
+minor_release=`echo $DISTRIB_RELEASE | cut -d . -f 2`
+
+if [ $major_release -lt 11 -o  $minor_release -lt 10 ]; then
+    echo "Oneiric 11.10 is the first supported release of the local provider"
+    exit 1;
+fi
+
+
+# Collect various status information about the system
+echo "#Local provider inspection"
+uname -a
+
+ifconfig
+virsh net-list --all
+
+ls /var/cache/lxc
+ls /var/cache/lxc/*
+
+lxc-ls
+
+# guess about the users data-dir
+if [ -n "${SUDO_USER}" ]; then
+    user=$SUDO_USER
+fi
+
+
+image=/var/lib/lxc/$image_name
+if [ -n "$image_name" -a -e "$image" ]; then
+    cat "$image/config"
+    chroot "$image/rootfs" bash -xc "
+        cat /etc/juju/juju.conf;
+        ls /usr/lib/juju/juju;
+        dpkg-query -s juju;
+        cat /etc/hostname
+        cat /etc/resolv.conf
+        cat /etc/hosts
+        tail -n 100 /var/log/juju/*.log
+    "
+fi
+
+
+
+
+
+