[Linux-cluster] Network failure results cluster environment unstable & fragile

Sat Feb 25 05:33:22 UTC 2006

Please help me to resolve my problem 

If network goes off on node1, and service which were not running on node1
 are started by node1 with shared storage mount point, which was already
 running on node 2 but both of these nodes are not  able to communicate to
 each other, node2 anyway already running the same service with shared
 storage mount point. Because of Fencing both of  these nodes try to kill
each other. Both of they got hanged up at "Stoping Cluster manager
Services.".In /var/log/messages, it shows fencing s1, fence successful. 

 If we disable fencing than 

 If network comes back nodes don't synchronize with each other. Shared
 storage mount point is available to both the servers. If they try to access
 storage at same storage gives IO errors. Hence this entire setup become
very unstable, fragile.

--- Deval kulshrestha
<deval.kulshrestha at progression.com> wrote:

> Hi 
> 
> I am struggling to get some help on following
> configuration. This setup is
> intended to put live in a data center for 24 x 7
> x365, any issue that makes
> my environment unstable is very critical here.
> 
> My HA Cluster Setup details
> 
> 1.	HP DL 360 G4p Server                       2nos.
> 2.	HP MSA 500 G2 (SAN)                     1nos.
> 3.	RedHat Enterprise Linux 4 ES  
> 4.	Red Hat Cluster Suite 4
> 
> 
> Server does have a HP SCSI HBA. MSA 500G2 is a scsi
> based SAN. Both of these
> server are connected to SAN using SCSI VHDCI cable.
> I used a network switch
> to establish network connectivity for the server.
> created a disk array of
> three HDD on SAN with two logical volumes than  I
> have installed RHEL 4
> Update 1 on both server(Servers are configured with
> RAID 1) than installed
> all HP drivers and management agents. After server
> configuration and OS
> installation I have installed Red Hat Cluster Suite
> v 4 on both the machine.
> 
>  
> 
> Than I have configured Cluster using Cluster
> Configuration Manager. Added
> member hosts, configured fence device and assigned
> to member host(HP iLO is
> certified as an fence device), Configured Failover
> domain with node
> priority, configured resources such as floating IP
> address, File System,
> Script, than configured service which need to be run
> in HA mode.
> 
>  
> 
> After configuring this I have tested with various
> scenario HA is working
> properly, when ever powered off any machine ,
> services fail over on
> available node. 
> 
> Problem:
> 
> 
> If network goes off on node1, and service which were
> not running on node1
> are started by node1 with shared storage mount
> point, which was already
> running on node 2 but both of these nodes are not
> able to communicate to
> each other, node2 anyway already running the same
> service with shared
> storage mount point. Because of Fencing both of
> these nodes try to kill each
> other. Both of they got hanged up at "Stoping
> Cluster manager Services.".In
> /var/log/messages, it shows fencing s1, fence
> successful. 
> 
> If we disable fencing than 
> 
> If network comes back nodes don't synchronize with
> each other. Shared
> storage mount point is available to both the
> servers. If they try to access
> storage at same storage gives IO errors. Hence this
> entire setup become very
> unstable, fragile.
> 
>  
> 
>  
> 
>  
> 
>  
> 
>  
> 
>  
> 
> With Regard
> 
> Deval
> 
> Progression Infonet Pvt. Ltd. 
> 55, Independent Electronic Modules, 
> Sector - 18, Electronic City, 
> Gurgaon - 122015
> 
> India
> Tel          : - 0124 - 2455070, Ext. 215, Fax:
> 91-124-2398647
> Mobile   : - 98186 -82509 
> URL        : - www.progression.com 
> 
>  
> 
>
===========================================================
> Privileged or confidential information may be
> contained
> in this message. If you are not the addressee
> indicated
> in this message (or responsible for delivery of the 
> message to such person), please delete this message
> and
> kindly notify the sender by an emailed reply.
> Opinions,
> conclusions and other information in this message
> that
> do not relate to the official business of
> Progression
> and its associate entities shall be understood as
> neither
> given nor endorsed by them.
>   
> 
>
-------------------------------------------------------------
> Progression Infonet Private Limited, Gurgaon
> (Haryana), India
> > --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
>
https://www.redhat.com/mailman/listinfo/linux-cluster

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

===========================================================
Privileged or confidential information may be contained
in this message. If you are not the addressee indicated
in this message (or responsible for delivery of the 
message to such person), please delete this message and
kindly notify the sender by an emailed reply. Opinions,
conclusions and other information in this message that
do not relate to the official business of Progression
and its associate entities shall be understood as neither
given nor endorsed by them.

-------------------------------------------------------------
Progression Infonet Private Limited, Gurgaon (Haryana), India

--
Linux-cluster mailing list
Linux-cluster at redhat.com
https://www.redhat.com/mailman/listinfo/linux-cluster

===========================================================
Privileged or confidential information may be contained
in this message. If you are not the addressee indicated
in this message (or responsible for delivery of the 
message to such person), please delete this message and
kindly notify the sender by an emailed reply. Opinions,
conclusions and other information in this message that
do not relate to the official business of Progression
and its associate entities shall be understood as neither
given nor endorsed by them.

-------------------------------------------------------------
Progression Infonet Private Limited, Gurgaon (Haryana), India