Linux, programming and more
Just imagine, your cluster is telling you if some resources getting stuck or manual interventions are required.
It could be a e-mail, push over notification or just a classic XMPP message.
This describes the basics of creating of own resources for your cluster.
The Corosync Cluster Engine is a Group Communication System with additional features for implementing high availability within applications.
A kind of masterpiece for high availability solutions and redundant systems with special requirement.
This stack is able to accept challenges in datacenter area to provide infrastructure high availability also across multiple different fire protection zones and over miles.
Sometimes if you add a new resource or change something in your cluster configuration you get some warnings like this one.
Or if you run a configuration syntax check:
1 | root@node1:~# crm configure verify |
1 2 3 4 5 6 7 | crm(live)configure# verify WARNING: mount_srv: default timeout 20s for start is smaller than the advised 60 WARNING: mount_srv: default timeout 20s for stop is smaller than the advised 60 WARNING: drbd_r0: default timeout 20s for start is smaller than the advised 240 WARNING: drbd_r0: default timeout 20s for stop is smaller than the advised 100 WARNING: pingnet: default timeout 20s for start is smaller than the advised 60 WARNING: pingnet: default timeout 20s for monitor_0 is smaller than the advised 60 |
Solution: define new operations, use the same times 240s and 100s
op start interval=”0″ timeout=”240s”
op stop interval=”0″ timeout=”100s”
Config snippet:
1 2 3 4 5 | primitive drbd_r0 ocf:linbit:drbd \
params drbd_resource="drbd0" \
op monitor interval="120s" \
op start interval="0" timeout="240s" \
op stop interval="0" timeout="100s" |
Really dirty fix (set global timeouts):
1 | crm configure property default-action-timeout="240" |
In this article I will explains how to set up and maintain an Active/Passive Cluster, using Pacemaker with Corosync with DRBD replication.
We have two nodes (same hardware), one active and another in passive mode. If the active node goes down, the passive one will automatically take its position and all running services.
This article was written complete for and with Debian 6-squeeze (With a lot of luck you could bring it to run on Ubuntu)
1. Basic notes
Node-1:
Hostname: node-1
IP: 192.168.2.101
Node-2:
Hostname: node-2
IP: 192.168.2.102
Partitioning:
/dev/vda1 System / ext4
/dev/vda5 reserved for DRBD – empty
/dev/vda6 swap area
Cluster IP: 192.168.2.100
We need this cluster IP to reach currently active node.
This is a physical network structure of the cluster:
[Read more…]