GitLab

 allow boot from partitions other than 1
 allow "-new" versions of freebsd/frisbee MFS

time it successfully mounts a filesystem.  Since we mount about 80 gazillion
filesystems in jails and these mounts are overlapped with other parts of the
setup process, it is quite likely that some jail's innocent use of
/usr/bin/foo is going to be blown out of the water by a random senseless HUP.

Grisly hack: when setting up jails, rename mountd.pid so mount can't find it.
Hah! hup THIS mount!

and in jails especially.

These get redirected to a file nowadays, so they shouldn't cause problems.

but now it is based on the topomap file (see tbsetup/gentopofile).
When the router type is static-ddijk, and there is a topomap file
present (gettopmap() returns valid) turn the topomap into input to
feed into the ddijkstra program. Take the output and feed that into
the parsing code.

NB: If we change the output of tmcd routing call, we have to change
the output of this program too. This will likely become a problem at
some point.

present (gettopmap() returns valid) turn the topomap into input to feed
into the genhostsfile program (see initial revision for os/genhostsfile),
Take the output and feed that into the parsing code.

NB: If we change the output of tmcd hostnames call, we have to change the
output of this program too. This will likely become a problem at some
point, although the format of the hostnames command has not changed for
about 2 years now.

The point of doing this is to remove the bottleneck of computing hostsfiles
from tmcd, which centrally computes them for every node, and in a big topo
(100s of nodes) that takes a lot of time and bw.

returns it as a perl data structure to the caller. The caller is
responsible for doing something meaningful with it.

Remove the static-ddijk code I added for the OSDI paper (revision
1.107), since the new topomap stuff supercedes that.

trying to copy it via NFS.

Add code to restart the tmcc proxy, which appears to die off for no
apparent reason. I now waitpid for it while the jail is running, and
restart if it exits.

At Mike's suggestion temporarily wrap tmcc proxy in a ktrace to see if we
can debug this problem. The ktrace file (using -a option) is written as
/var/emulab/jails/nodeid/tmcc.ktrace.

issue route commands when the INET is undefined, as it will be for
physical interfaces used for veths.

interfaces when configuring veth interfaces. Instead, tmcd now returns
actual configuration information for those interfaces, albeit slightly
modified to make it apparent what their role is. See revision 1.232 of
tmcd/tmcd.c for more information.

See initial revision for tbsetup/gentopfile for more details. Note
that a gzipped version of the file is also created by gentopofile, and
we try to get that first to cut down on the amount of data copied over
(the topomap is highly compressable).

separately for now. This is in prepratation for computing hosts files
on the nodes instead of in tmcd.

changes yet, so backing them out. Previous commit message is meaningless.

for physical interfaces underlying veth interfaces; The interfaces are
configured as part of regular ifconfig, rather then as a hack in veth
configuration.

Most of the info return is left blank, especially INET (ip address)
which indicates the special handling.

pounding /user too much (via NFS of course).

from os_setup.

Needs to be backed out later.

the pnode rebooted, the file got multiple entries causing the generated
gated config to have syntax errors.

	- increase IP interrupt queue size from 50 to 128
	- turn on magic nfs-permission-denied-avoidance sysctl

the dijkstra program on the linkmap in the tbdata directory, and feed
those results into the pattern match loop as if the results came from
tmcd.

Obviouse problem; if we change the format that is returned from tmcd.

is addressed in tmcc.c

socket files does not exists (connect returns ENOENT) retry until it
appears. The caller is responsible for timeout (via -t option) just as
for the tcp case.

in the timestamps, and seeing what was taking so much time.

* Increase the overlap from two to four seconds (have vnodesetup exit
  earlier) so that the next jail can start sooner. Using getrusage it
  seemed pretty clear that there were plenty of cycles available to
  start the next jail going a little earlier. Instead of looking for
  the watchdog pid file to appear, I touch (in mkjail) a specific file
  in the jail directory , and have vnodesetup look for that file
  instead of the pid.

* Cut back the number of inodes in the filesystem! Sheesh, we ere
  creating 137,000 inodes for a 50MB vnode filesystem. Takes 1
  second to do that. Reduced to 30,000 and the time when to .1
  seconds. Could probably go even lower.

* Add -p option to fsck when restoring a jail from an existing
  filesystem. No need to force check the filesystem if its clean.

* Instead of running MAKDEV inside each jail filesystem (which was
  also taking .8 seconds to do), do it just once and stash a copy in
  /var/emulab/jails. Each jail does a "hier cp" of that master copy
  into the jail filesystem. Reduced the time to .05 seconds.

So, for the results:

Original time to boot a jail: 8.2 seconds, with the last two seconds
overlapped with the next one.

Original time to reinstate a jail after halting it (filesystem left
behind): 6.5 seconds, with the last 1.5 seconds overlapped with the next
one.

Original time to boot 16 jails on a pc850: 2 minutes and 12 seconds.

---------------------------------------------------------------------------

New time to boot a jail: 6.2 seconds, with the last 4 seconds overlapped
with the next one.

New time to reinstate a jail after halting it: 6 seconds, with the last 4
seconds overlapped with the next onw.

New time to boot 16 jails on a pc850: 88 seconds. Does this number ring any
bells? :-)

tbsetup/libtestbed.pm.in which we use on boss. I have modified these
routines in two ways (which need to be merged back into original
version at some point).

* import/require Time::HiRes module (/usr/ports/devel/p5-Time-HiRes)
  to get higher resolution gettimeofday() function that reports
  microseconds and seconds. Change the output routine to append the
  microseconds to the original HH:MM:SS timestamps that get printed.

* When turning timestamps on dynamically with TBDebugTimeStampsOn(),
  set an environment variable, and read that in so children also know
  to turn them on.

late.  So wait 5 seconds and try again if the first tmcc call fails.

XXX this is a hack that should be replaced!