GitLab

Otherwise, the daemon cannot write it!

If the constant MAXWRITEBUFMEM was set to zero (unlimited), imageunzip
would ignore any attempts to limit the size of writebuf memory.

So this happened twice this week. Two Utah Emulab users, starting a
sliver at the Utah Emulab CM. Basically, while Flack was starting up a
sliver for the user, they decided to log into the web interface and
recreate their encrypted certificates. So they register a slice, and
Flack gets the slice certificate. Then the user changes their
certificate on Emulab. The CM notices that the certificate in the
slice credential and the certificate the user presented are
different. For a nonlocal user we want to update our record (dubious
on its own), but for a local user we really do not want to do
anything.

Users do wacky things.

Too many to change. Instead make the hack to watchdog as well.

For example: "tmcc -- ntpdrift -40.224"

over, we can now just pull it with anon git.

close the saved STDOUT/STDERR descriptors until the new ones are open
and dup'ed into fileno 1 and 2, and the LOG descriptor closed. This
was causing SelfLoader to get confused; it has a wacky test that is
supposed to look for a fork and reopen the handle on __DATA__, so that
parent and child have its own fd. Not sure I really understand the
sequence of events though.

mistakes while doing so).

them via XMLRPC. Must do this by hand on the web interface until
fixed.

them via XMLRPC. Must do this by hand on the web interface until
fixed.

Create "getstorageconfig" call in libsetup, following the tradition with
other tmcc information fetching routines.  It's guts were yanked out of
rc.storage.

Outside of calling the moved code from libsetup, rc.storage was also
changed slightly to store the returned information using "Storable"
since it's now dealing with an array of hashes instead of raw lines
of output from tmcc.

Keep gcc happy.

Did this for frisbee, but forgot here.

For the day when we configure local storage...

In order to hook in via the "generic vnode" path for setting up
blockstores under FreeNAS, I've done a bit of shuffling in order to
make things more OS-independent and reusable.

* mkvnode.pl

Moved to clientside/tmcc/common.  OS-dependent bits (really only some
IPtables stuff) abstracted, and moved to tmcc/linux/libvnode.pm.

* libvnode.pm

Moved generic vnode stuff to a new module.  Moved miscellaneous
utility functions to a new module.  Left OS-specific stuff.  Not
really sure if what is left should be merged into libsetup/liblocsetup
or left here - deferring this decision for now.

* libgenvnode.pm

New module currently containing generic vnode stuff.  Currently, the
VNODE_* predicates are here.

* libutil.pm

New module containing miscellaneous utility functions (fatal,
mysystem, mysystem2, setState, etc.)

Files referencing libvnode.pm have been updated, as have the relevant
Makefiles.