It is a good discipline to read machine logfiles periodically and be on the lookout for messages that stand out.

One recent linux message that might start appearing on a Linux 64 bit machine concerns HPET ( High Precision Event Timer ) and here is an example:

CE: hpet increasing min_delta_ns to 15000 nsec

Now running off and web searching for results is probably a longer way round than reading the next paragraph or two, which have the important things you might want to know first.

Firstly this is a Linux kernel feature designed to help power efficiency so on the whole it is a positive thing.

It is neither brand new nor some serious bug.

It can be disabled if you are bored of reading about it already ( hpet=disable )

There is an Intel presentation titled ‘Getting maximum mileage out of tickless’ which has on page 3 a gentle introduction to why hpet is a desirable feature.

For the more technically adept there are some further details here and here.

Back to the message I was seeing in my logs…there are bug reports regarding messages of the sort 'hpet increasing min_delta_ns', in particular, if you are seeing lots of these messages in your logs, then it might be worthwhile engaging with the kernel development team and sharing your output.

For me this laptop install is a brand new Xubuntu 64 bit install and is still in the process of being fully configured. When I have all my applications running as I wish then I’ll check the logs again and see if things look okay. With just a single message appearing occasionally, it might be that waiting a few months for the Lucid upgrade is (and the newer kernel 2.6.32) will be the path of least effort.

For folks running Debian / Ubuntu servers in the cloud, this information regarding tickless might be worth a read.

Advertisements