[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[Linux-cluster] cluster name length bug




Hi, while I was trying to setup a new cluster I've found a bug in the valid name length of a cluster. The MAX_CLUSTER_NAME_LEN is 16, but if I put a cluster name of 16 chars when a node tries to connect to the cluster the node is refused to connect and in 'dmesg' you can see this:
CMAN: attempt to join with cluster name 'qmail-be-cluster' refused
CMAN: attempt to join with cluster name 'qmail-be-cluster' refused
CMAN: attempt to join with cluster name 'qmail-be-cluster' refused
CMAN: attempt to join with cluster name 'qmail-be-cluster' refused
Searching through the code I found that It was here where the problem exists:
-----------------------------------------------------------------------------
               /* Check the cluster name matches */
               if (strcmp(cluster_name, joinmsg->clustername)) {
                       printk(KERN_WARNING CMAN_NAME
": attempt to join with cluster name '%s' refused\n",
                              joinmsg->clustername);
                       return -1;
               }
----------------------------------------------------------------------------

The name of the cluster was the same in all nodes: qmail-be-cluster.
It's easy to reproduce the problem. Just create a cluster with a name of 16 chars and then It's impossible to connect to the cluster.

This is my conf file in all nodes:

<?xml version="1.0"?>
<cluster config_version="13" name="qmail-be-cluster">
<fence_daemon clean_start="1" post_fail_delay="0" post_join_delay="3"/>
       <clusternodes>
               <clusternode name="qmail-be-01" votes="1">
                       <fence>
                               <method name="1">
<device name="manopla" nodename="qmail-be-01"/>
                               </method>
                       </fence>
               </clusternode>
               <clusternode name="qmail-be-04" votes="1">
                       <fence>
                               <method name="1">
<device name="manopla" nodename="qmail-be-04"/>
                               </method>
                       </fence>
               </clusternode>
               <clusternode name="qmail-be-05" votes="1">
                       <fence>
                               <method name="1">
<device name="manopla" nodename="qmail-be-05"/>
                               </method>
                       </fence>
               </clusternode>
       </clusternodes>
       <cman/>
       <fencedevices>
               <fencedevice agent="fence_manual" name="manopla"/>
       </fencedevices>
       <rm>
               <failoverdomains/>
               <resources/>
       </rm>
</cluster>


I'm using Fedora Core 4 with this rpm's :

kernel-smp-2.6.14-1.1653_FC4
GFS-6.1.0-3
GFS-kernel-smp-2.6.11.8-20050601.152643.FC4.20
cman-kernel-smp-2.6.11.5-20050601.152643.FC4.18
cman-1.0.0-1
ccs-1.0.0-1
dlm-kernel-smp-2.6.11.5-20050601.152643.FC4.17
dlm-1.0.0-3
fence-1.32.1-1
rgmanager-1.9.34-5
lvm2-2.01.08-2.1
lvm2-cluster-2.01.09-3.0

We need to put this on bugzilla, but I don't know how to do it. :(

Thanks
German Staltari


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]