Hi,

On Tuesday 13 February 2018 05:03 PM, Chetan Sharma wrote:

If i execute a LXC Job. If this job execution failed due to any reason, then in that case LXC-stop is not called, directly lxc-destroy is called by which LXC instance is still running on system and consumes memory and Space.

It would be helpful if you can specify the version of lava-dispatcher in which you observe this behavior.

I say this because over the past few months we have made improvements to the code that takes care of most of the corner cases in this regard, some of them are listed below:

* https://projects.linaro.org/browse/LAVA-1091 * https://projects.linaro.org/browse/LAVA-1227 * https://projects.linaro.org/browse/LAVA-984

But still there are few non-reproducible rare occurrence of such uncleaned LXCs as explained in https://projects.linaro.org/browse/LAVA-1162

How we can can make sure that this instance destroys with Job completion either Failed/Passed/Incomplete.

For now it has to be done manually.

We have a work-in-progress to handle this as seen in https://projects.linaro.org/browse/LAVA-1146

If this could be re-produced consistently on your environment and have the bandwidth to code a patch, then we encourage to come up with such a patch to solve this issue which is much welcomed.

Thank You.