CentOS 5 Server :: GFS With Little 'Cluster' As Possible / No CLVM
Mar 15, 2010
What I did not realize was, that DLM uses the external Ethernet Interface even when talking to the local machine/node. So iptables was blocking my DLM daemon. With iptables down or the TCP port for DLM opened, cman starts, mount works.What I have here is a fibrechannel SAN which will be directly attached to several servers in the near future. Thise servers should be enabled access to a single filesystem on the SAN (shared).I heard that the right filesystem choice for this kind of setup would be GFS, because it has a Distributed Lock Manager and one FS journal for each node.
But I am having trouble setting up GFS. I have managed to create a GFS on a small testvolume (local HDD so far), but am unable to mount it. It seems that GFS/DLM needs a lot of cluster services to run, which I do not all understand / know how to correctly setup. Also: Will the lock_dlm stuff need Ethernet communications to handle file locks? And if so, will it fetch the node list from /etc/cluster/cluster.conf to determine who to talk to?
I created a cluster with two nodes and a machine for managers with luci, if a machine reboot the cluster function by transferring the resource (IP address), if forced to stop the machine (pull the plug) the cluster does not work.
I'm having some trouble configuring clustering in a 2-node cluster, with no shared FS. Application is video streaming, so outbound traffic only...The cluster is generally ok - if I kill -9 one of the resource-applications, the failover works as expected. But it does not failover when I disconnect the power from the service owning node (simulating a hardware failure). clustat on the remaining node shows that the powered-down node has status "Offline", so it knows the node is not responding, but the remaining node does not become the owner, nor start up the cluster services/resource-applications. eth0 on each node is connected via a crossover cable for heartbeat, etc. Each eth1 connects to a switch.
I have a two node cluster, and a third system which has luci installed.
node1 is nfs0 node2 is nfs1
both nodes have identically the same configuration. They have a fresh installation of Centos 5.5 + yum update. I am unable to join nfs1 to the cluster, as it is giving me the following issue:
Sep 29 23:28:00 nfs0 ccsd[6009]: Starting ccsd 2.0.115: Sep 29 23:28:00 nfs0 ccsd[6009]: Built: Aug 11 2010 08:25:53 Sep 29 23:28:00 nfs0 ccsd[6009]: Copyright (C) Red Hat, Inc. 2004 All rights reserved.
I have just installed a two server cluster with ricci luci and conga on centos 5.6 32bit , both servers are vmware guests and have a shared storage disk connected to them both
with a GFS2 file system on them + fencing agents configured to work with VMware Vcenter.
(this is supported by vmware and works great on 4 other centos clusters i have been runing for 4 monthes with no CLVMD).
In this setup i used for the first time CLVMD as recommnded by RedHat so i could have the flexablitly of LVM under the GFS2 file system but , i have been getting some Strange problem with it , some times after a developer has done some IO heavy task like unziping a file or a simple TAR the load goes to 10 - 15 and no task can be killed , trying to reboot the server hangs.
After hard shutting the server every thing works ok until the next time some one does the same IO work as before.
I'm having a very strange problem with red hat cluster. After testing it in the lab, I tried to install a new cluster and received the following error: cman not started: Cannot start, cluster name is too long or other CCS error /usr/sbin/cman_tool: aisexec daemon didn't start I've checked the internet for that error, but nothing. I decided to to take the example from cluster.conf man, which looks like that :
[Code]...
And still I get the same error. I can find my servers both from DNS (FQDN and short name) and they also appear in /etc/hosts.
I just wanna simulate HPC and and other kinds of clusters in VMware workstation 7.0 , in my HP 520 laptop which is dualcore and 3Gb RAM. So can u please help me out regarding this.. I am interested to work in clusters. I am new to this hpc and other clusters..please can any one gve me document on cluster installation and configuration. I would be grateful to them.. I am using Centos OS.
I am researching the CentOS Cluster Setup. Does anyone know of a guide on this or have a thread linked that I can look at? I want to build a small cluster and then be able to add to it as it grows. Mainly I want to have web hosts, data, and mail behind it. The DNS will probably be on there own.
I have added apache application into redhat cluster. If the apache process failed, it will cause the service group fail over. I want to disable the application monitoring. I search throung Internet and it depends on the apache init script. If I change /etc/init.d/httpd status always return 0(success). THe service group will not fail over even apache failed.
However, I want to remain the application monitoring but stop failover. If the application like apache failed, I want Redhat cluster to try restart the application like 3 or 4 times. If restart failed, it will not failover.
I am trying to do the cluster storage with the Rock Cluster storage operating system . I have install the rock cluster on the main server . and I want to connect the client with the PXE boot . when I starting the client then it boot from the PXE boot into the mode of compute node. but it ask the where is to be stored the I have give the path..then server says that not able to find the directory path.
Steps:- 1) insert-ethers 2) client is started with the PXE boot. 3) it detects the dhcp . 4) At last it demand where is to be started by cd-rom, harddisk ,nfs etc.
then I started with the nfs and I have given the lanIp of server.and server is detecting the client but client is not finding the filesystem directory. into export partition directory but it is not taking path.
PATH:- /export/rock/install it is not finding this path then it is not able to start the o/s from PXE boot..Is there any solution and manual of rock or anyother solution then u can reply to me...
well sorry I'm new in this but I just made my cluster with about 4 good machines each with 2 or more gbs of memory and dual-core processor but the question is i can install an operating system in nodes as having the frontend or not different if possible, and how i can do it because i don't now how and i don't find material and tutorials that can help me..
I'm currently using 2 virtual machines with opensuse 11.2 (virtualbox). The network and the cluster is working fine, running /etc/init.d/openais start the cluster goes up and:
Running 'zypper install lvm2-clvm' results in a message: Resolving package dependencies... Problem: nothing provides libcpg.so.2()(64bit) needed by lvm2-clvm-2.02.45-16.2.x86_64 Solution 1: do not install lvm2-clvm-2.02.45-16.2.x86_64 Solution 2: break lvm2-clvm by ignoring some of its dependencies So it cannot find the cpg library, which appears to be a part of openais (if I believe google). I have configured the following repositories:
[Code]....
What could be the problem here? Am I just missing a repository or is cluster LVM not supported on opensuse 11.3?
I am trying to build GFS2 cluster with 2 or 3 Fedora 14 nodes, but I've encountered some problems from the start. First luci does not work at all in Fedora 14. There is no luci_admin and even if I manage to start luci service, I get a blank white screen when I try to open it from the browser. I've googled a bit and found that I'd might be able to setup GFS if I manage to build cluster.conf manually and start the cluster suite, but I cannot find documentation on how to create cluster.conf anywhere. If anyone knows how to setup GFS2 without a cluster suite or how to configure cluster.conf.
i'm a familiar with Linux environement ( fedora 10 user ) and i got a project in a training where i have to create a cluster with two nodes where i have to set up a number of VMs that will run applications such as ( Samba, Ldap, Zimbra, ...) but i don't know how to virtualize on top of a cluster ! i would like to know how that can be done, and how is it possible to let the VMs get ressources ( RAM & CPU ) from the two nodes ??
I am working in a project that needs to set up an Apache Web Cluster. The cluster needs to be High-availability (HA) cluster and Load-balancing cluster. Someone mentioned the use of Red Hat Cluster Suite, but, honestly, I can't figure out how it works, I haven't been able to configure it correctly. The project currently have two nodes, but we need to support at least three nodes in the cluster.
I have lack of understanding of CentOS in general. I have looked for a remedy on other forums and google, but haven't been able to find the answer. I have a 3 node cluster that was functioning great until I decided to go offline for awhile. My config is as follows: node 2: vh1 node 3: vh2 node 4: vh6 All nodes connect to a common shared area on an iscsi device (vguests_root)
Currently vh2 and vh6 connect great, however since putting the machines back online I can no longer connect with vh1. A dmesg command on vh1 reveals the following: GFS2: fsid=: Trying to join cluster "lock_dlm", "Cluster1:vguest_roots" GFS2: fsid=Cluster1:vguest_roots.2: Joined cluster. Now mounting FS... GFS2: fsid=Cluster1:vguest_roots.2: can't mount journal #2 GFS2: fsid=Cluster1:vguest_roots.2: there are only 2 journals (0 - 1) .....
I have an annoying problem with a 2 nodes cluster that I don't understand.
One of my server is totally shutdown.
When I start cman after a reboot on the other server, it hangs on "Starting fencing...". The process which is actually hanging is : /sbin/fence_tool -w -t 300 -m 45 join
When I start this program manually, I get this output :
# /sbin/fence_tool -w -t 300 -m 45 join Waiting for all 2 nodes to be members Waiting for all 2 nodes to be members Waiting for all 2 nodes to be members Waiting for all 2 nodes to be members (and so on...)
What I don't understand is why the fence_tool program wait for the 2nd node to join the cluster and don't fence the other node.
Using google with search option: cman not started: Can't find local node name in cluster.conf /usr/sbin/ cman_tool: aisexec daemon didn't start.I found this URL...I have found the config_version in cluster.conf. Unfortunately, as everyone may have noticed, english is not my native tongue so I am having trouble understanding the part "Make sure you bumped the cluster config version number". Can anyone enlightened me on what should I be doing so that I could "bump" the cluster config version?
i am using the redhat cluster suite (luci and ricci) on my centos 5.3. i have 2 nodes in a cluster. when i poweroff the first node on wich a vm service is running, the service switchtes to node2. so far, so good :) but when i restart node1 the service is not failback to node1! i have created a failover domain with both nodes and priorized whre node1 has prio1 and node2 has prio2.
This is my first post, and i have a question with the fence software for VMWare ESXi. The fence_vmware agent only works with ESX, and redhat (in you GIT repository) has submited a new agent called fence_vmware_ng that claims to work with ESXi. But the problem is that they do not specify the version that works with that. Anybody has test the fence_vmware_ng agent for VMWare ESXi 4.0 ?, i follow the instructions here: [URL]...and i can install the software, the API from VMWARE site, etc, but in the moment when i run the agent nothing happen, The agent connects to server, i see in logs, but the off-reboot-on operations not works. Only works status operations, that return the state of a virtual machine. I have CentOS 5.3 (fully today updated) with RHCS.
I put together a 3 PC cluster approx. 6 years ago with Redhat, at that time I was in Grad. School. Now I am in a small company and we are planning to go for a cluster setup. I have chosen to go with Centos5.
The last time around I used NFS, RSH etc. for file sharing, communication etc. Would it be OK for me to go with the same stuff or some new software/technology has come in.
The Centos cluster setup is for high-performance computing.
I have made a cluster between two server.In luci I can see that my cluster is green and the two nodes to.I have make an IP resource and associate it to a service : green : I can relocate the service from a node to the other one and the IP appears in the list of IP addresses The problem is that I have made the same in order to configure tomcat and postgresql and it does not work...I put my configuration only for ip and tomcat:
I am working on the beginning of implementing a two-node cluster with shared storage (GFS) and IP address. Both machines are virtual on VMware ESX 3.5, that should not make a difference, but that is the background.current status is that I have a single node cluster built with only the IP address configured within the cluster. The issue that I am having is that I have configured a service to contain only the IP address resource, however, when I go into cluster management that "service" does not register. As such, I cannot bring it online, ping it, etc. below is my cluster.conf configuration:
I have problems setting up a (2 server) cluster. I get the following messages in syslog:
Oct 14 13:02:21 korzel ccsd[7823]: Starting ccsd 2.0.115: Oct 14 13:02:21 korzel ccsd[7823]: Built: Sep 3 2009 23:26:21 Oct 14 13:02:21 korzel ccsd[7823]: Copyright (C) Red Hat, Inc. 2004 All rights reserved. Oct 14 13:02:21 korzel ccsd[7823]: cluster.conf (cluster name = cl_vpanel, version = 1) found.
[code]....
Tried with the "basic" hosts file as well, eg with only the localhost lines in it. Both server run centos 5.3, up-to-date.There is no real error anywhere which might explain WHY aisexec daemon won't start. When i start openais manually (/etc/init.d/openais start) it starts without any errors.
I'm trying to configure an ISCSI/DRBD high-availability cluster and I'd like to know what is the best option between OpenAIS and Heartbeat. I've seen they both are included in Centos Repos, yet OpenAIS requires installing 2 addition repos to install Pacemaker (EPEL and Clusterlab repos).
