[Linux-cluster] ricci is very unstable in one nodes

[Ben Turner]

There is an issue with ricci timeouts that was fixed recently:

https://bugzilla.redhat.com/show_bug.cgi?id=564490

I'm not sure but you may be hitting that bug.  Symptoms include: luci isn't able to get the status from the node, timeouts when querying ricci, etc.  The fix should be released with 5.6

On the mysql service there are some options that you need to set.  Here are all the options available to that agent:

mysql
Defines a MySQL database server

Attribute	Description
config_file	Define configuration file
listen_address	Define an IP address for MySQL server. If the address is not given then first IP address from the service is taken.
mysqld_options	Other command-line options for mysqld
name	Name
ref	Reference to existing mysql resource in the resources section.
service_name	Inherit the service name.
shutdown_wait	Wait X seconds for correct end of service shutdown
startup_wait	Wait X seconds for correct end of service startup
__enforce_timeouts	Consider a timeout for operations as fatal.
__failure_expire_time	Amount of time before a failure is forgotten.
__independent_subtree	Treat this and all children as an independent subtree.
__max_failures	Maximum number of failures before returning a failure to a status check.

If I recall correctly you may need to tweak:

shutdown_wait	Wait X seconds for correct end of service shutdown
startup_wait	Wait X seconds for correct end of service startup

There can be problems relocating the DB if it takes too long to start/shutdown.  If you are having problems relocating with luci it may be a good idea to test with:

# clusvcadm -r <service name> -m <cluster node>

-Ben



----- "fosiul alam" <expertalert at gmail.com> wrote:

> Hi
> I have 4 nodes cluster,
> It was running fine. but today one nodes is giving trouble
> 
> From luci Gui interface, when i try to relocate service into this node
> and trying to relocate from this nodes to another nodes
> 
> from luci gui interface, its showing :
> 
> Unable to retrieve batch 1908047789 status from
> beaver.domain.local:11111: clusvcadm start failed to start httpd1:
> Starting cluster service "httpd1" on node "http1.domain.local" -- You
> will be redirected in 5 seconds.
> also
> 
> The ricci agent for this node is unresponsive. Node-specific
> information is not available at this time. :
> 
> but ricci is running on problematic node ,
> ricci 7324 0.0 0.1 58876 2932 ? S<s 14:40 0:00 ricci -u 101
> 
> there is not any firewall running.
> 
> iptables -L
> Chain INPUT (policy ACCEPT)
> target prot opt source destination
> 
> Chain FORWARD (policy ACCEPT)
> target prot opt source destination
> 
> Chain OUTPUT (policy ACCEPT)
> target prot opt source destination
> 
> Chain RH-Firewall-1-INPUT (0 references)
> target prot opt source destination
> 
> port 11111 is runningg
> 
> netstat -an | grep 11111
> tcp 0 0 0.0.0.0:11111 0.0.0.0:* LISTEN
> 
> 
> but still ricci is very unstable , and i cant relocate any service on
> this node or i cant relocate any service away from this node.
> 
> from problematic node if i type this
> 
> clustat
> Cluster Status for ng1 @ Thu Sep 23 20:24:02 2010
> Member Status: Quorate
> 
> Member Name ID Status
> ------ ---- ---- ------
> beaver.xxx.local 1 Online, rgmanager ::: luci is running from this
> server
> publicdns1.xxxx.local 2 Online, rgmanager
> http1.xxxx.local 3 Online, Local, rgmanager
> mail01.xxxxx.local 4 Online, rgmanager
> 
> Service Name Owner (Last) State
> ------- ---- ----- ------ -----
> service:httpd1 mail01.xxxx.local started
> service:mysql-server http1.xxxx.local started ------------------- this
> is the problematic node
> service:public-dns publicdns1.xxxxxx.local started
> 
> I cant move that service mysql-server from this node or cant relocate
> any service on this node ..
> I am very confused.
> 
> what shall i do to fix this issue ??
> 
> thanks for your advise.
> 
> 
> 
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster