On 03/24/2009 02:29 PM, David Lehman wrote:
On Tue, 2009-03-24 at 14:21 -1000, David Cantrell wrote:
Looks ok to me.
Is this the patch you were talking about to get grow support working
better right now? I would eventually like this functionality to live in
pyparted or libparted.
It is. We really want also to teach libparted to stop inserting the
cylinder-alignment constraint so we can also start using
maximizePartition again, as this is not a complete solution.
What this will do is correct our initial allocations so that
maximizePartition won't be as likely to grow the first partition it
seems to be huge, leaving the others as they are. This one will get us
perhaps 80% of the way to a reasonable use of the available space in
these weird scenarios.
Sounds reasonable for now.
I would love to see the majority of anaconda's partition allocation and
growing logic moved into pyparted.
Hopefully we can get there.
Dave
On 03/24/2009 01:10 PM, David Lehman wrote:
parted.disk.Disk.maximizePartition leads to libparted applying
a cylinder-alignment constraint, which we would frikking ask for
if we wanted it. So, for now, we can use this to identify chunks
of free space that will initially be allocated to requests whose
max size will prevent them from making full use of their share
of the space.
Cases where this comes into play are ones where an initially
larger request has a relatively low max size and an initially
smaller request has no max size at all. Since we base the
allocation of free space for growing on the requests' initial
size, we could previously end up with very large chunks of
unused free space.
---
storage/partitioning.py | 30 ++++++++++++++++++++++++++++++
1 files changed, 30 insertions(+), 0 deletions(-)
diff --git a/storage/partitioning.py b/storage/partitioning.py
index f716657..9f42a72 100644
--- a/storage/partitioning.py
+++ b/storage/partitioning.py
@@ -835,11 +835,41 @@ def growPartitions(disks, partitions):
log.debug("growable total is now %d sectors" % disk_total)
# now we loop through the partitions...
+ # this first loop is to identify obvious chunks of free space that
+ # will be left over due to max size
+ leftover = 0
+ limited = {}
+ unlimited_total = 0
+ for part in growable:
+ # calculate max number of sectors this request can grow
+ req_sectors = part.partedPartition.geometry.length
+ share = float(req_sectors) / float(disk_total)
+ max_grow = (share * disk_free)
+ max_sectors = req_sectors + max_grow
+ limited[part.name] = False
+
+ if part.req_max_size:
+ req_max_sect = (part.req_max_size * (1024 * 1024)) / sectorSize
+ if req_max_sect< max_sectors:
+ mb = ((max_sectors - req_max_sect) * sectorSize) / (1024*1024)
+
+ log.debug("adding %dMB to leftovers from %s"
+ % (mb, part.name))
+ leftover += (max_sectors - req_max_sect)
+ limited[part.name] = True
+
+ if not limited[part.name]:
+ unlimited_total += req_sectors
+
+ # now we loop through the partitions...
for part in growable:
# calculate max number of sectors this request can grow
req_sectors = part.partedPartition.geometry.length
share = float(req_sectors) / float(disk_total)
max_grow = (share * disk_free)
+ if not limited[part.name]:
+ leftover_share = float(req_sectors) / float(unlimited_total)
+ max_grow += leftover_share * leftover
max_sectors = req_sectors + max_grow
max_mb = (max_sectors * sectorSize) / (1024 * 1024)
log.debug("%s: base_size=%dMB, max_size=%sMB" % (part.name,
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
--
David Cantrell <dcantrell@xxxxxxxxxx>
Red Hat / Honolulu, HI
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
