I’ve just pushed a set of trivial modules system scripts that will hopefully ease your deployment of Cloudera Distribution for Hadoop 3 Beta 4 on your university cluster… partly, at least. This sad “partly” made me think about the current state of things on IT and HPC.

Over time I’ve learnt that there are several unexpected issues when deploying hadoop on custom clusters that you don’t own. Those are mainly related to software management policies, non-root access (being auto-deployment unfriendly), quotas and queueing or “batch” systems.

Ignoring most of these “fixable” issues, it becomes apparent that the most juicy problem for a sysadmin trying to get the most of hadoop-related tools is the batch system. Be it SGE, SLURM or other non-DRMAA compliant exotic batch system implementations, you’ll have to deal with annoying integration quirks at some point, granted.

Making it all work can be challenging to say the least… but the question is: does it have to be that hard ?

Continue reading →

So we have a simple migration at hand looking at the current virtualization landscape, right ?

1. Use virt-convert from python-virtinst. In other words: OVF to libvirt’s XML
2. Launch KVM with the resulting migrated files (both images and metadata)
3. Do trivial configurations inside the guest machine to match host environment

That was my idealized view. It turns to be this way:

1. Try virt-convert to discover that it fails when converting from OVF.
2. Try to fix libvirt python code libraries (OVF XML parser).
3. When the code is fixed, the resulting metadata files are malformed.
4. Try with the commercial “vCenter Converter Standalone”: ERROR: “OVF contains multiple virtual machines”.
5. Download ~500MB of “VMWare Server” to just use the “vmware-vdiskmanager” binary (1.4MB).
6. Construct libvirt’s xml, with some hand-edited sections.
7. virt-manager fails when connecting via VNC to your remote VM due to a Ubuntu-CentOS cross-distro issue involving NetCat parameters. Place a workaround and report the bug.
8. Optimize both virtual network and disk performance switching to VirtIO.
9. Rebuild guest ramdisk image in order to load VirtIO drivers in the boot process.

Want more details ? Keep reading…
Continue reading →