Open Stack

On 2015-12-20 12:35:34 -0800 (-0800), Clark Boylan wrote:
> Looking at the dstat logs for a recent fail [0], it did help in that
> more memory is available. You now have over 1GB available but still less
> than 2GB.
[...]

As Clark also pointed out in IRC, the workers where this is failing
lack a swap memory device. I have a feeling if you swapoff before
this point in the job you'll see it fail everywhere.

For consistency, we probably should make sure that our workers have
similarly sized swap devices (using a swapfile on the rootfs if
necessary).
-- 
Jeremy Stanley