From: Jim Prescott (jgp_at_harn.ceas.rochester.edu)
Date: 28 Jul 2005 13:53:21 -0400
Some questions on growfs in Solaris 8 and up. I'm not using Disksuite
or LVM, just normal /dev/rdsk/c#t#d#s# partitions.
I start with 2 physically adjacent partitions on the same disk; the
first has a UFS partition, the one after it is unused. I use format
to increase the size of the UFS partition so it also includes all of
the unused partition. I then use:
growfs -M /mountpoint /dev/rdsk/c2t0d0s0
Q1 Was all of that safe to do on a mounted, active partition? I know
that growfs write-locked the fs while working and I could shorten
the locked interval by doing multiple growfs commands with -s.
Q2 How do the newfs options given to growfs relate to the newfs options
originally used? Eg if we had used -c or -i on the original newfs
do I need to use the same options again on the growfs?
Q3 If the partition is strictly for files >1M but was newfs'd without
specifying -i, is it ok to add "-i 300000" to the growfs? Would the
new value only affect the new cylinder groups?
Q4 When running growfs it says it writes super-block backups to dozens
of locations within the original fs. Are these locations always going
to be the same locations used by the original newfs or are there some
changes that could occur between newfs and growfs (which could span years)
that would result in growfs scribbling superblocks into random files?
A nice enhancement for growfs would be the ability to tell it to grow
into the full partition but do it in N steps, sleeping X seconds after
each step. You can do that now with -s but this is tedious and error
prone, exactly the kind of thing you'd want a program to do.
Thanks for any help you can provide.
-- Jim Prescott - Computing and Networking Group firstname.lastname@example.org School of Engineering and Applied Sciences, University of Rochester, NY