COAL issue related to SAPI-280

[nerakhon]

I've just been deploying manta to my coal and found out that the slow I/O within the VM affects the imagadm action simply causing it to take more than an hour which is made the default timeout of SAPI and causes failure of the manta-deploy-coal.

Would you consider adding some tunable for COAL? Or is there a plan to fix this in manta?

[jclulow]

@nerakhon Ideally we would solve this, but in the meantime there is a documented workaround for the import of the large Marlin compute image: https://joyent.github.io/manta/#workaround-for-manta-marlin-image-import-failure

We apologise for the inconvenience!

[nerakhon]

This actually doesn’t help as the local unpacking takes that long, not the download.

[jclulow]

@nerakhon Oh. As an additional step, you can run imgadm import $uuid manually in the global zone of each compute node where you need to create marlin zones. Once that has completed, the creation of marlin zones from the image shouldn't take long at all.