View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0001931||CentOS-5||ricci||public||2007-04-17 18:07||2007-11-12 03:47|
|Product Version||5.0 - i386|
|Target Version||Fixed in Version||5.1|
|Summary||0001931: Ricci module "ricci-modrpm" crashes, preventing Cluster configuration from Luci|
|Description||When using "luci" to configure a new Cluster, "luci" displays the following error:|
"Please be patient - this cluster's configuration is being modified.
Creating node XXX for cluster YYY
A problem occurred when installing packages: failed to locate/execute module"
This is apparently the result of "luci" asking "ricci" to run "/usr/libexec/ricci-modrpm", which crashes with:
terminate called after throwing an instance of 'std::basic_string<char, std::char_traits<char>, shred_allocator<char> >'
This prevents any Cluster configuration using "luci", the preferred method.
Also, the Upstream Provider's logo, trademark and links still exist on every "luci" web page.
|Additional Information||To use "luci", check the "Clustering" option in the CentOS 5 Installer, then run:|
|Tags||No tags attached.|
I just wanted to confirm this issue.
It makes Luci quite useless.
I would also like to confirm that this happens on Centos 5.0 x86_64 as well.
There is also a problem with system-config-cluster tool of Centos 5 . Its opened as a bug in upstream vendor's bugzilla. In /usr/share/system-config-cluster/CommandHandler.py, path for cman_tool is wrongly given as /sbin/cman_tool when its actually /usr/sbin/cman_tool. Because of this,we can't administer a cluster with this utility. Solution is to create a symlink and things will work fine . Please see the following link for details.
What I have found is in CentOS 5 atleast, the same issue (wrong path for cman_tool) is there in /var/lib/luci/Extensions/CommandHandler.py. But unlike the bug in system-config-cluster , creating a symlink is not solving the problem. I am pretty surprised that bugs like this have not been detected in beta period especially with upstream vendor.
This is what i did to fix this issue:
mv /etc/redhat-release /etc/redhat-release.orig
echo "Red Hat Enterprise Linux Server release 5 (Tikanga)" > /etc/redhat-release
Fixed my problem.
This is is also confirmed on X86_64.
(0005188) and (0005317) allowed me to use luci successfully to create/manage my new cluster.
Fix for (0005317) is rolled into 5.1
(0005188) seems to no longer be required, since system-config-cluster has been fixed to look at /usr/sbin/cman_tool
|2007-04-17 18:07||richs||New Issue|
|2007-04-17 18:07||richs||Status||new => assigned|
|2007-04-23 04:51||rainigraf1||Note Added: 0004925|
|2007-05-17 09:42||rhysmcmurdo||Note Added: 0005183|
|2007-05-18 11:59||nidal||Note Added: 0005188|
|2007-06-07 10:41||YonatanE||Note Added: 0005317|
|2007-06-08 19:58||daugustus||Note Added: 0005329|
|2007-06-20 15:40||danieldk||Status||assigned => acknowledged|
|2007-11-12 03:firstname.lastname@example.org||Status||acknowledged => assigned|
|2007-11-12 03:email@example.com||Note Added: 0006267|
|2007-11-12 03:firstname.lastname@example.org||Status||assigned => resolved|
|2007-11-12 03:email@example.com||Fixed in Version||=> 5.1|
|2007-11-12 03:firstname.lastname@example.org||Resolution||open => fixed|