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

[Linux-cluster] Re: Linux-cluster Digest, Vol 46, Issue 48



linux-cluster-request redhat com wrote:
Send Linux-cluster mailing list submissions to
	linux-cluster redhat com

To subscribe or unsubscribe via the World Wide Web, visit
	https://www.redhat.com/mailman/listinfo/linux-cluster
or, via email, send a message with subject or body 'help' to
	linux-cluster-request redhat com

You can reach the person managing the list at
	linux-cluster-owner redhat com

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Linux-cluster digest..."


Today's Topics:

   1. RE: is clvm a must all the time in a cluster ? (Roger Pe?a)
   2. Re: gfs_tool: Cannot allocate memory (Bob Peterson)
   3. Ethernet Channel Bonding Configuration in RHEL	Cluster Suite
      Setup (Balaji)
   4. R: [Linux-cluster] Ethernet Channel Bonding Configuration in
      RHELCluster Suite Setup (Leandro Dardini)
   5. probe a special port and fence according to the	result (Peter)
   6. Re: probe a special port and fence according to	the result
      (Lon Hohberger)
   7. Re: Ethernet Channel Bonding Configuration in RHEL	Cluster
      Suite Setup (Lon Hohberger)
   8. Re: probe a special port and fence according to	the	result
      (Brian Kroth)
   9. Re: probe a special port and fence according to the	result (Peter)
  10. Re: Redundant (multipath)/Replicating NFS (isplist logicore net)
  11. Re: Redundant (multipath)/Replicating NFS (gordan bobich net)
  12. Re: Redundant (multipath)/Replicating NFS (isplist logicore net)
  13. Re: gfs_tool: Cannot allocate memory (rhurst bidmc harvard edu)


----------------------------------------------------------------------

Message: 1
Date: Thu, 28 Feb 2008 22:00:04 -0500 (EST)
From: Roger Pe?a <orkcu yahoo com>
Subject: RE: [Linux-cluster] is clvm a must all the time in a cluster
	?
To: linux clustering <linux-cluster redhat com>
Message-ID: <233538 63310 qm web50602 mail re2 yahoo com>
Content-Type: text/plain; charset=iso-8859-1


--- Roger Peña <orkcu yahoo com> wrote:

  
--- Steffen Plotner <swplotner amherst edu> wrote:

    
Hi,

You asked below: can I run the cluster with GFS
without clvmd? The answer is yes. I believe in
having the least number of components running, and
found that clvmd had start up problems (and then
refuses to stop) after doing an update of RHEL4
during December.  

      
I apply those updates this month so I guess I am
seeing the same as you 

    
It is clearly possible to use GFS directly on a
      
SAN
    
based LUN. 
      
I know that but, as you already said, the problem is
"having a uniq name for the filesystem to all the
nodes"

    
The trick of course is how to deal with
the /dev/sdb reference which will probably not be
the same on all hosts. To fix that use udev rules
that provide a symlink (using the serial number of
the LUN, for example) by which the GFS file system
can be referred to in /etc/fstab.
      
I am sure udev rules works but definitly seting that
enviroment is more complex that use a LV :-)
so, what happen is I use the shared LV just as a
local
LV?
each node will treated the same way as it treat the
LV
from the local disks. I guess that will not be a
problem as far as I do not work with the VG
metadata,
am I right?




    
We have converted 2 clusters in the past few
      
months
    
because we have had real problems with clvmd
misbehaving itself during startup. At this point
      
it
    
is a pleasure to let the cluster boot by itself
      
and
    
not to have to worry about GFS file systems not
being mounted (ccsd, cman, fenced, iscsi, gfs).
      
not activating LV even when clvmd is running? it
happen to me several times in the last month
;-) 
that is why I want to get rid of lvm :-)
    
                                   ^^^
I mean clvm

sorry, I want to use lvm, but I dont want to use clvm
all the time, just when, very rarely, need to create
or resize shared lv

cu
roger

__________________________________________
RedHat Certified ( RHCE )
Cisco Certified ( CCNA & CCDA )


      Get a sneak peak at messages with a handy reading pane with All new Yahoo! Mail: http://ca.promos.yahoo.com/newmail/overview2/



------------------------------

Message: 2
Date: Thu, 28 Feb 2008 21:03:36 -0600
From: Bob Peterson <rpeterso redhat com>
Subject: Re: [Linux-cluster] gfs_tool: Cannot allocate memory
To: linux clustering <linux-cluster redhat com>
Message-ID: <1204254216 3272 15 camel technetium msp redhat com>
Content-Type: text/plain

On Thu, 2008-02-28 at 10:28 -0500, rhurst bidmc harvard edu wrote:
  
ioctl(3, 0x472d, 0x7fbfffe300)          = -1 ENOMEM (Cannot allocate 
    

Hi Robert,

The gfs_tool does most of its work using ioctl calls to the gfs kernel
module.  Often, it tries to allocate and pass in a huge buffer to make
sure it doesn't ask for more than the kernel needs to respond with.

In some cases, it doesn't need to allocate such a big buffer.
I fixed "gfs_tool counters" for a similar ENOMEM problem with
bugzilla record 229461 about a year ago.  (I don't know if that bug
record is public or locked so you may not be able to access it, which
is out of my control--sorry).

I should probably go through all the other gfs_tool functions, including
the two you mentioned, and figure out their minimum memory requirements
and change the code so it doesn't ask for so much memory.

Perhaps you can open up a bugzilla record so I can schedule this work?
Also, you didn't say whether you're on RHEL4/Centos4/similar, or
RHEL5/Centos5/similar.

Regards,

Bob Peterson
Red Hat GFS




------------------------------

Message: 3
Date: Fri, 29 Feb 2008 17:58:00 +0530
From: Balaji <balajisundar midascomm com>
Subject: [Linux-cluster] Ethernet Channel Bonding Configuration in
	RHEL	Cluster Suite Setup
To: ClusterGrp <linux-cluster redhat com>
Message-ID: <47C7FA50 8080202 midascomm com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Dear All,

    I am new in RHEL Cluster Suite.

    I have Configure Cluster and Rebooted the system and then cluster 
become active in primary node and other node as passive and member 
status becomes Online for both the cluster nodes
 
    In Cluster Suite i am monitoring the resources as scripts files and 
ipaddress and During network failure one of the node or both the nodes 
are removed from the cluster member and All the resources are stopped 
and then rebooted the system only both the system are joining into the 
cluster member.

    I have followed the RHEL Cluster Suite Configuration document 
"rh-cs-en-4.pdf" and I have found out Ethernet Channel Bonding in Each 
Cluster Nodes will avoid the network single point failure in cluster 
system.
 
    I have configured the Ethernet Channel Bonding with mode as 
active-backup without fence device.

    Ethernet Channel Bonding Configuration Details are
    1. In " /etc/modprobe.conf" file added the following bonding driver 
support  
       alias bond0 bonding
       options bonding miimon=100 mode=1
    2. Edited the "/etc/sysconfig/network-scripts/ifcfg-eth0" file added 
the following configuration
       DEVICE=eth0
       USERCTL=no
       
       MASTER=bond0
       SLAVE=yes
       BOOTPROTO=none
    3. Edited the "/etc/sysconfig/network-scripts/ifcfg-eth1" file added 
the following configuration
       DEVICE=eth1
       USERCTL=no
       
       MASTER=bond0
       SLAVE=yes
       BOOTPROTO=none
    4. Created a  network  script for the bonding device 
"/etc/sysconfig/network-scripts/ifcfg-bond0"
        DEVICE=bond0
        USERCTL=no
        
        IPADDR=192.168.13.109
        NETMASK=255.255.255.0
        GATEWAY=192.168.13.1
    5. Reboot the system for the changes to take effect.
    6. Configure Ethernet Channel Bonding
    7. Rebooted the system and then cluster services are active on both 
the nodes and member status of current node is Online and other node as 
Offline

    I need clarification about Ethernet Channel Bonding will work with 
Fence Device or not.

    I am not sure why this is happening. Can some one throw light on this.

Regards
-S.Balaji




       




















------------------------------

Message: 4
Date: Fri, 29 Feb 2008 13:51:05 +0100
From: "Leandro Dardini" <l dardini comune prato it>
Subject: R: [Linux-cluster] Ethernet Channel Bonding Configuration in
	RHELCluster Suite Setup
To: "linux clustering" <linux-cluster redhat com>
Message-ID:
	<6F861500A5092B4C8CD653DE20A4AA0D511B91 exchange3 comune prato local>
Content-Type: text/plain;	charset="iso-8859-1"

 

  
-----Messaggio originale-----
Da: linux-cluster-bounces redhat com 
[mailto:linux-cluster-bounces redhat com] Per conto di Balaji
Inviato: venerdì 29 febbraio 2008 13.28
A: ClusterGrp
Oggetto: [Linux-cluster] Ethernet Channel Bonding 
Configuration in RHELCluster Suite Setup

Dear All,

    I am new in RHEL Cluster Suite.

    I have Configure Cluster and Rebooted the system and then 
cluster become active in primary node and other node as 
passive and member status becomes Online for both the cluster nodes
 
    In Cluster Suite i am monitoring the resources as scripts 
files and ipaddress and During network failure one of the 
node or both the nodes are removed from the cluster member 
and All the resources are stopped and then rebooted the 
system only both the system are joining into the cluster member.

    I have followed the RHEL Cluster Suite Configuration 
document "rh-cs-en-4.pdf" and I have found out Ethernet 
Channel Bonding in Each Cluster Nodes will avoid the network 
single point failure in cluster system.
 
    I have configured the Ethernet Channel Bonding with mode 
as active-backup without fence device.

    Ethernet Channel Bonding Configuration Details are
    1. In " /etc/modprobe.conf" file added the following 
bonding driver support  
       alias bond0 bonding
       options bonding miimon=100 mode=1
    2. Edited the "/etc/sysconfig/network-scripts/ifcfg-eth0" 
file added the following configuration
       DEVICE=eth0
       USERCTL=no
       
       MASTER=bond0
       SLAVE=yes
       BOOTPROTO=none
    3. Edited the "/etc/sysconfig/network-scripts/ifcfg-eth1" 
file added the following configuration
       DEVICE=eth1
       USERCTL=no
       
       MASTER=bond0
       SLAVE=yes
       BOOTPROTO=none
    4. Created a  network  script for the bonding device 
"/etc/sysconfig/network-scripts/ifcfg-bond0"
        DEVICE=bond0
        USERCTL=no
        
        IPADDR=192.168.13.109
        NETMASK=255.255.255.0
        GATEWAY=192.168.13.1
    5. Reboot the system for the changes to take effect.
    6. Configure Ethernet Channel Bonding
    7. Rebooted the system and then cluster services are 
active on both the nodes and member status of current node is 
Online and other node as Offline

    I need clarification about Ethernet Channel Bonding will 
work with Fence Device or not.

    I am not sure why this is happening. Can some one throw 
light on this.

Regards
-S.Balaji

    

When there is a network failure, each member cannot reach the other one. Each member trigger the fence script to turn off the other one. Unfortunately the network is off, so the low level network interface take the fence script connection request in mind and wait for the network to come up. When the network is again available, each member can reach the fencing device and turn off the other. There is no simple way to avoid this. You can make network near 100% fault proof using bond device or you can use a STONITH fencing device, so the first member who regain network prevents the other to fence, or you can use three members. In a network failure no one have the quorum to fence the others.

Leandro



------------------------------

Message: 5
Date: Fri, 29 Feb 2008 15:05:23 +0100
From: Peter <p elmers gmx de>
Subject: [Linux-cluster] probe a special port and fence according to
	the	result
To: linux clustering <linux-cluster redhat com>
Message-ID: <99FC135E-6367-41B5-B6B5-AA1FEC5D9833 gmx de>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes

Hi!

I am planning to test a service for availability and if the service is  
not available anymore, fence the node and start the service on another  
node. Therefore, i need a clue how to probe a port with system tools.
Sure, with telnet there is a possibility to check. But i have no idea  
to avoid the interactivity. So it is too difficult for me to implement  
and usually not allowed in this environment.

On Solaris, there is the "ping -p" command to test the availability of  
a application waiting on a port. But for Redhat, i have no clue.

Could you please help?


Peter



------------------------------

Message: 6
Date: Fri, 29 Feb 2008 09:20:23 -0500
From: Lon Hohberger <lhh redhat com>
Subject: Re: [Linux-cluster] probe a special port and fence according
	to	the result
To: linux clustering <linux-cluster redhat com>
Message-ID: <20080229142023 GD6571 redhat com>
Content-Type: text/plain; charset=us-ascii

On Fri, Feb 29, 2008 at 03:05:23PM +0100, Peter wrote:
  
Hi!

I am planning to test a service for availability and if the service is not 
available anymore, fence the node and start the service on another node. 
Therefore, i need a clue how to probe a port with system tools.
Sure, with telnet there is a possibility to check. But i have no idea to 
avoid the interactivity. So it is too difficult for me to implement and 
usually not allowed in this environment.

On Solaris, there is the "ping -p" command to test the availability of a 
application waiting on a port. But for Redhat, i have no clue.

Could you please help?
    

Net Cat? (man nc ) ?

  
Dear All,
Ping is working but the members are not in cluster and its forming a new cluster at both the nodes
The following Messages are logged in "/var/log/messages"
CMAN: forming a new cluster
CMAN: quorum regained, resuming activity


Regards
-S.Balaji


then cluster services are active on both 
the nodes and member status of current node is Online and other node as 
Offline


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