feat: add support for powervs #20
Conversation
This however requires to bypass the osuosl ppc hack.
nikromen
force-pushed
the
powervs-support
branch
from
e4342e5 to
936edee
Compare
| # When this flag exists, the osuosl hack in enable-swap.sh will be SKIPPED | ||
| RUN if [ "$POWERVS" = "true" ]; then \ | ||
| echo "Running on PowerVS" > /etc/copr-builder/powervs-enabled; \ | ||
| fi |
There was a problem hiding this comment.
This is weird, do we really need to build an OCI image for ppc64le twice, once for powervs & once for hypervisors (and the rest)?
Seems like we could invent some form of runtime check.
There was a problem hiding this comment.
we have hardwired hack for swap for power9 and 10 but it is only for osuosl openstack since we have (according to the comment there) only one volume for the image... I need to bypass this for powervs since we are allocating additional volume (as we do with the rest of VMs)
There was a problem hiding this comment.
I missed your reply; we should discuss this on some meeting; what I mean here is that it would be nice to detect the powervs machiens at machine boot time, not at image build time. That would be the way to keep one OCI image for all the ppc64le machines (hypervisors, osuosl and powervs).
IOW, this check should be done in enable-swap (I'm renaming it to copr-partitions.sh).
|
/retest |
| # about the "pool ID" (== particular hypervisor). | ||
| generic_mount | ||
| elif grep -E 'POWER9|POWER10' /proc/cpuinfo; then | ||
| elif grep -E 'POWER9|POWER10' /proc/cpuinfo && ! test -f /etc/copr-builder/powervs-enabled; then |
There was a problem hiding this comment.
what about
elif test -f /etc/copr-builder/powervs-enabled; then
generic_mount
elif ...2 participants
This however requires to bypass the osuosl ppc hack.