GitLab

all free blocks before throwing out ones that are too small (-F).  There
were a fair number of cases where a small free chunk was adjacent to a larger
one and we were tossing out the smaller.  This does increase the risk that
we will run out of memory building the free list.  If that happens, we can
make an incremental cleanup pass.

when linktest fails on swapin path. Also, add a file link to the
swaplog file, which still contains the linktest output.

Added test for emulab synchronization mechanism.

the swap log otherwise.

wait_all, just add one to the error_count instead of the exit code,
otherwise sync gets unhappy.

fixes and the ns comparison dialog is improved a little.

the kernel via a magic ioctl (see shd/shd.h).  For example:

	imagedump -C /dev/shdXX /usr/testbed/images/FBSD410-STD.ndz

Would build a list of allocated ranges for the given image and pass them
to /dev/shdXX via an ioctl.

Document the -D option to choose the hash.

I got side-tracked on this while looking for reasons why a simple delta image
is so large.  So I thought I would reevaluate our choice of 64 as the default
value.  Bottom line is that it is a pretty good choice.

directly.

the port and retry with a different address/port if so.

Also, check for immediate failure of the TBBackground call so that we can
return at error if it fails (most likely because someone forgot to do a
post-install after boss-install and it cannot write the indicated logfile)

and return a distinct error when we fail to get the port

These files get regenerated on every boot anyway
(i.e., rc.sysinit removes them and they keep coming back!)

Note: I don't spend all my time analyzing the RHL boot sequence, I noticed
this as a consequence of looking at what changed in a "delta" disk image
created with the new imagezip.

number.

the sql cluestick, so the queries shouldn't be as bad as before.

ltmap file so that it does the proper calculation of delay, bandwidth,
and loss, as it is done in assign_wrapper. More specifically, the
delay/bandwidth/loss between two nodes in a lan is a function of nodea
delay/bandwidth/loss, and nodeb rdelay/rbandwidth/rloss, where the "r"
values are for the "from the lan to the node" direction.  These might
very well be different then the other values for asymmetric links.

rebuilt. Seems to be a moving target ...

special stuff at Utah (MAINSITE=1).

* Before aging out, copy the entries to another DB on boss so that we
  have it forever!

* As an extra special hack of the century, copy those entries over to
  ops and store in a DB over there for Kevin to look at. This code
  will go away at some point.

not getting updated if you bring in a source tree with older timestamps.

case things are under maintenance.