Just a quick hint for those of you who are experiencing problem with getting node-agents up and running in a Glassfish cluster.
If the problem is that the asadmin start-agent command simply times out/uses a long time to start, and even possibly returning a filenotfoundexception. The problem could be as simple as a firewall blocking traffic to/from the master-node. A quick way to check it is to simply disable the FW.
Note that if the master-node does not see the node-agent status, it is also caused by FW-issues, fix the FW and then restart the node-agent(s), then it all should work like a charm!
A nice way of rebooting a frozen Linux-based server when you do not have a power switch available is to use the SysRq-abilities. (Ubuntu Example.) Blackbox does not have a default macro installed for just that, but you can create a new macro and insert the following:
This will create a working macro for writing any data in cache to disk, kill all processes (Sigkill), remount mounted disks and perform a reboot.
Sometimes the backspace button can be a real pain on SLES-systems, but the fix is simple!
Add this to ~/.vimrc:
I have been playing around with GPFS on SLES for some days now and have compiled a quick HOWO for installation of the system.
- You must have root-access for the servers you are installing on
- You must compile GPFS kernel modules
- You must have access to the GPFS-packages from IBM
- Install all 4 rpm files
- Install needed packages for compilation: zypper in gcc xorg-x11-devel gcc-c++ kernel-source
- Download the latest patch from IBM*
- Compile it: cd /usr/lpp/mmfs/;make Autoconfig;make World;make InstallImages
Now you should have GPFS up and running without too much hazzle.
(* Patches are not always needed to make it work, but in case you encounter compilation errors, the latest patch should fix them)