GitLab

pcvm nodes until a better fix is decided.

reloads. Usage is:

	sched_reload <options> -t pctype [pctype ...]

In other words, schedule a reload for nodes of a particualar type (or
types) like pc600, pc850, etc.

One I actually did the math (duh) 64 or even 48 is way too high.
16 works much better for the static case.

Minor nits.

Another set of changes to make the script clean up logs left aroung by prior
runs (skipped due to timeout waiting for close).

Small change to continue when one of several logfiles is not closed before
the script times out on it (rather than exiting immediately).  Can potentially
leave behind old logs, but these will be picked up by subsequent runs.

real nodes get. Also, run a proper os_select on jailed nodes, *after*
the os for the physical node is setup, since otherwise stated will not
be happy.

Fixes for dealing with failed os_load. Previously, if os_load would
fail, os_setup would wait for those nodes anyway since it had no idea
what nodes had failed (and we do not want to just quit from os_setup
since that might cause a lot of extra power cycles). Now, for each
node that got an os_load, check its eventstate; it should be in ISUP
immediately after os_load exits (since thats what os_load waited for),
and if its not, then mark that node as failed. Note though that failed
loads no longer result in the node going into hwdown, since 99 percent
of the time its a busted user image, not a hardware problem. I figure
we will catch real hw errors via the reload daemon, when it sends
email about nodes not finishing.

Do not bother with doing the vnode setup if any of the phys nodes
failed to setup. Leads to cascading errors and prolongs the angony by
another few minutes. Might revisit this later.

Remove local WaitTillAlive() function, and switch to using the version
I put into libdb a couple of weeks ago.

Fix up a bunch of print statements to be nicer.

Simple command-line interface to the ready bits. Its primary
purposes are:

* Manually report ready for nodes that can't do it themselves
* Get a list of which nodes are ready, so that you can figure out
  which one(s) aren't reporting in
* Clear ready bits so you can use them again without restarting the
  experiment
* Make it possible to poll ready bits on boss/ops

calculation based on the size of the image file. Okay, to avoid all
you folks from going to see what bit of dreck I came up with, here it
is:

    my $sb     = stat($imagepath);
    my $chunks = $sb->size / (1024 * 1024);
    $maxwait   = int((($chunks / 100.0) * 25) + (4 * 60));

Note the replacement of one hardwired number (15) with several dozen
new ones!

I like it anyway, cause I hate waiting 2*15 minutes when a 60 second
load fails.

as the op_mode it is currently running. If not, force it and send mail.

This fixes the "stuck-in-reloading" phenomenon we occasionally see when
the states get messed up. Now anytime it loads PXEFRISBEE it should force
the mode to RELOAD, and it will stop reloading the first time it hits
RELOADDONE.

and osid in the partitions table, use that, but, if not, use the
one from the os_info table. Previously, only the osid from the
partitions table was used.

then remove special case for sending REBOOTING event in node_reboot/power
when using NORMAL mode. Now SHUTDOWN is always sent. (Important side note:
SHUTDOWN needs to be a valid state in every machine now.)

(the IXPs) with more.  Will probably put this check back in later,
with the maixumum number of links coming from the database, per
node type.

script name (basename, not full path).
Also moved the "use" lines below the package declaration, as I think
they are supposed to be.

builds properly on Linux.

on linux. Add a makefile ifdef for Linux/Freebsd that sets up the
proper set of programs. Note that the new image I'm making will have
the event libraries installed!

in terms of not putting them in the reloadpending experiment, etc.

frisbee.redux makefile, not the imagezip makefile.  Besides making
more sense, this ensure that all the frisbee client objects get built
with the same compiler (the imagezip/unzip code will be built with gcc30
if imagezip NTFS support is enabled)

percent, and optimize for space. Prelude to creating smaller jails
on local nodes, as soon as I can get SFS running inside a jail the way
I want it (in which case users will have access to their project and
home dirs on the file server).

Add Mike's IPADDR change, with slight modification. tmcd will specify
a list of ip addresses as a comma separated list, which are converted
to -i options to pass to jail. Kernel will restrict bind to these IPs.