Good evening list, I'm having a rather confusing issue with 4 new nodes that I put up. JNA doesn't seem to want to work in production even though it works on my test VM that should have pretty much the same setup.
I have created a series of setup scripts to bring up new nodes. I am using a tested and tried configuration that has been running very well in production for a while and in all cases the following holds true: OS: CentOS6 with latest updates 2.6.32-431.17.1.el6.x86_64 Java: Java(TM) SE Runtime Environment (build 1.6.0_30-b12) Cassandra: 1.0.12 JNA: Tried both jna-3.5.1.jar, and jna-4.1.0.jar In both my old production servers and my test setup (which I can recreate over and over and get the same results), startup yields: "CLibrary.java (line 109) JNA mlockall successful" However, for some reason on the NEW production boxes, I get: "Unable to link C library. Native methods will be disabled" The JNA libs are exact copies and part of our setup scripts. There are only 2 major differences I can think of. The first is the hardware. All are 64 bit platforms, but the newer servers has 48 GB of ram and newer processors than the older boxes and my test vm. I don't think that would have anything to do with it, but it's worth mentioning. Both production sets are Xeons though. The second difference is that the new image would've been set up by Rackspace while my test uses an older image and brings it up to date with yum. However, as part of setup I apply yum update to all of them, so they should be running at least the same versions of whatever they have in common. Another interesting note is that while looking for "jna" on the disk, I also see what looks like an ELF binary in /tmp named /tmp/jna-1073564104/jna4691935553862497129.tmp. Thank you ahead of time for any help you may be able to offer. -- David Daeschler
