Aaron Klingaman [Fri, 24 Mar 2006 19:42:07 +0000 (19:42 +0000)]
regen
Aaron Klingaman [Fri, 24 Mar 2006 19:22:23 +0000 (19:22 +0000)]
more changes to make it fit in new MA terminology
Marc Fiuczynski [Fri, 24 Mar 2006 04:45:15 +0000 (04:45 +0000)]
Support "other storage device" PCI class.
Fix to search for drivers that match the full PCI_ID first and then look
for drivers that use PCI_ANY for subdevice and subvendor ids.
Aaron Klingaman [Fri, 24 Mar 2006 00:52:53 +0000 (00:52 +0000)]
more rewording and reorg. to fit into the arch series documents
Aaron Klingaman [Thu, 23 Mar 2006 21:45:12 +0000 (21:45 +0000)]
resize
Mark Huang [Tue, 21 Mar 2006 14:57:29 +0000 (14:57 +0000)]
- run createrepo as well
- if run under sudo, allow user to delete the headers/ and repodata/
directories
Aaron Klingaman [Fri, 17 Mar 2006 00:39:28 +0000 (00:39 +0000)]
first take at new booting nodes pdn, very largely based off of existing
tech doc for bootmanager. still needs bootcd doc integrated
Aaron Klingaman [Fri, 17 Mar 2006 00:36:07 +0000 (00:36 +0000)]
sequence diagram (original ppt and export png) for use in new booting nodes pdn
Mark Huang [Fri, 10 Mar 2006 18:20:34 +0000 (18:20 +0000)]
- run yum-arch on Bootstrap RPMS directory before using
Marc Fiuczynski [Fri, 10 Mar 2006 06:32:21 +0000 (06:32 +0000)]
Updated to work with new merge_hw_tables.merge_files() functionality.
Marc Fiuczynski [Fri, 10 Mar 2006 06:31:07 +0000 (06:31 +0000)]
Rewrote merge_files function to handle case when multiple modules claim
they support a partical PCI device. For example, the 8139cp and 8139too
modules both claim they can handle the generic RTL 8139 eth chipset.
Also incorporate support to distinguish when a driver can only handle
specific subvendor/subdevice IDs. E.g., the bcm5700 driver currently
does not handle all broadcom devices---some must still be handled by
tg3 (e.g., on the Dell PowerEdge 850).
Marc Fiuczynski [Fri, 10 Mar 2006 06:25:14 +0000 (06:25 +0000)]
Added optional filename argument to write_modprobe_config() function for
testing purposes.
Aaron Klingaman [Tue, 31 Jan 2006 21:19:03 +0000 (21:19 +0000)]
implement two changes to better handle, identify, and recover from
situation where no detected network driver is found. this should prevent
machines from installing, and booting, only to become unresponsive due to
no loaded network driver.
the two changes are:
1. write out modprobe.conf at each boot, instead of only after install
2. if no network devices are written to modprobe.conf during boot,
send a message to the tech contacts about missing net driver, and
put the node into debug.
Aaron Klingaman [Mon, 23 Jan 2006 18:29:06 +0000 (18:29 +0000)]
when checking for usable disks, also skip disks starting with dm-
(device mapper, ie lvm)
Aaron Klingaman [Mon, 23 Jan 2006 18:12:10 +0000 (18:12 +0000)]
bump release number
Aaron Klingaman [Mon, 23 Jan 2006 18:10:57 +0000 (18:10 +0000)]
filter out partitions from block devices by using device minor number,
not partition name format
Aaron Klingaman [Thu, 12 Jan 2006 21:43:12 +0000 (21:43 +0000)]
move code block to write out /etc/planetlab/session to ChainBootNode, since
it being in UpdateNodeConfiguration only is called after the first
boot. POD should work immediately after install now.
Mark Huang [Tue, 20 Dec 2005 18:59:23 +0000 (18:59 +0000)]
- bump version number for new release
Aaron Klingaman [Tue, 20 Dec 2005 05:56:06 +0000 (05:56 +0000)]
patch from E Jaffe to fix eth0 config files with mac address (if specified)
Mark Huang [Thu, 15 Dec 2005 16:50:39 +0000 (16:50 +0000)]
support pldistro "variants"
Aaron Klingaman [Fri, 2 Dec 2005 01:40:22 +0000 (01:40 +0000)]
patch from Marc F to optionally pull in extra kexec args from /kargs.txt
Aaron Klingaman [Mon, 21 Nov 2005 17:46:43 +0000 (17:46 +0000)]
update and rebuild docs to reflect readding the api server in the configuration
and new search path for node configuration file
Aaron Klingaman [Mon, 21 Nov 2005 17:25:07 +0000 (17:25 +0000)]
before looking in /usr and /usr/boot for the node configuration file,
try the root file system (/plnode.txt)
Aaron Klingaman [Mon, 21 Nov 2005 17:15:17 +0000 (17:15 +0000)]
- back out change that was going to be used for federation stage 1, which
was to assume boot server = api server, returning api server configuration
variable.
- bump version
Aaron Klingaman [Wed, 16 Nov 2005 19:57:44 +0000 (19:57 +0000)]
*** empty log message ***
Aaron Klingaman [Wed, 16 Nov 2005 19:13:43 +0000 (19:13 +0000)]
update to include where the source is, and misc. updates to match
the implementation
Mark Huang [Wed, 26 Oct 2005 15:32:10 +0000 (15:32 +0000)]
- bump version number to reflect fixes
- fix old style boot server file parsing
- fix alpina-BootLVM.tar.gz
Mark Huang [Wed, 26 Oct 2005 15:30:57 +0000 (15:30 +0000)]
- aaron seems to have included a few other binaries in the last manually
built BootLVM tarball
Mark Huang [Wed, 26 Oct 2005 15:26:15 +0000 (15:26 +0000)]
for the old style boot server file, that contains multiple servers,
read only the first line
Mark Huang [Mon, 3 Oct 2005 14:34:58 +0000 (14:34 +0000)]
- use a valid node_id
Mark Huang [Fri, 23 Sep 2005 01:23:18 +0000 (01:23 +0000)]
- add back 1.5
Mark Huang [Fri, 23 Sep 2005 01:22:29 +0000 (01:22 +0000)]
- add back 1.12
Mark Huang [Fri, 23 Sep 2005 01:14:26 +0000 (01:14 +0000)]
- XXX back out 1.5 temporarily for a tag
Mark Huang [Fri, 23 Sep 2005 01:11:37 +0000 (01:11 +0000)]
- bump version to 3.1.9
- XXX add back BOOT_API_SERVER temporarily for a tag
Mark Huang [Fri, 23 Sep 2005 01:10:59 +0000 (01:10 +0000)]
- make rpm version number consistent
Mark Huang [Thu, 22 Sep 2005 22:28:41 +0000 (22:28 +0000)]
- bump release number to reflect initrd and new-tarball fixes
Mark Huang [Thu, 22 Sep 2005 22:27:43 +0000 (22:27 +0000)]
- read the primary kernel version from the /boot/initrd-boot symlink,
rather than the first entry in /lib/modules
- ...thus building the initrd for the correct kernel, in case there are
orphaned modules in /lib/modules installed by, say, vnet
Mark Huang [Thu, 22 Sep 2005 22:26:50 +0000 (22:26 +0000)]
- first try downloading the new PlanetLab-Bootstrap.tar.bz2, then fall
back to the old alpina-BootstrapRPM.tar.bz2
Mark Huang [Thu, 22 Sep 2005 22:25:35 +0000 (22:25 +0000)]
- skip the yum bootstrap if told to by InstallBootstrapRPM
Mark Huang [Thu, 22 Sep 2005 22:25:13 +0000 (22:25 +0000)]
- get_system_modules: allow kernel_version to be passed in
Mark Huang [Thu, 22 Sep 2005 22:24:28 +0000 (22:24 +0000)]
- restore the InstallBase step, which will be skipped if the new
all-inclusive PlanetLab-Bootstrap.tar.bz2 is unpacked
Aaron Klingaman [Wed, 21 Sep 2005 20:29:52 +0000 (20:29 +0000)]
rather than code in the name of the boot server into the boot manager
configuration, pull it off of the boot cd. this change requires that
an api server be running on the boot server.
Mark Huang [Thu, 8 Sep 2005 20:06:03 +0000 (20:06 +0000)]
deprecate alpina-BootstrapRPM and the InstallBase step; instead just unpack PlanetLab-Bootstrap and go
Mark Huang [Wed, 7 Sep 2005 18:15:46 +0000 (18:15 +0000)]
- bump release number to beat .planetlab.planetlab
Mark Huang [Wed, 7 Sep 2005 18:15:31 +0000 (18:15 +0000)]
- fix release number
Mark Huang [Sat, 3 Sep 2005 21:44:16 +0000 (21:44 +0000)]
Solve the bootstrap problem by including any just built packages in
the yum configuration. This cooperates with the PlanetLab build
system.
Mark Huang [Sat, 3 Sep 2005 21:41:43 +0000 (21:41 +0000)]
- don't look for binaries or build debug package
Mark Huang [Fri, 2 Sep 2005 20:24:58 +0000 (20:24 +0000)]
deprecate the BootstrapRPM tarball, build and install all other tarballs until v2 cds are out of circulation
Mark Huang [Fri, 2 Sep 2005 19:51:17 +0000 (19:51 +0000)]
- build bootmanager
Mark Huang [Fri, 2 Sep 2005 19:09:58 +0000 (19:09 +0000)]
- script to build node tarball
Mark Huang [Fri, 2 Sep 2005 19:09:33 +0000 (19:09 +0000)]
- build PlanetLab-Bootstrap.tar.bz2 (new, full version of alpina-Bootstrap.tar.bz2)
Aaron Klingaman [Fri, 5 Aug 2005 22:18:13 +0000 (22:18 +0000)]
for each boot, after the vserver net configuration files ae updated,
write out the new session value in /etc/planetlab/session.
Aaron Klingaman [Fri, 29 Jul 2005 17:38:22 +0000 (17:38 +0000)]
rollback previous regex change, attempts to mount base device should
still happen. instead, make sure it fails correctly. also, a few spelling
mistakes
Aaron Klingaman [Fri, 29 Jul 2005 15:53:14 +0000 (15:53 +0000)]
fix regex that identifies which device to attempt to mount for removable
devices during configuration file searching. before, it was returned
the base device name (ie sda) as a possible device to mount. it should only
return <device>[0-9]+$
Aaron Klingaman [Fri, 29 Jul 2005 15:40:02 +0000 (15:40 +0000)]
be way more verbose in what is happening during configuration file
searching/parsing
Aaron Klingaman [Thu, 28 Jul 2005 23:23:47 +0000 (23:23 +0000)]
allow any node that has their node_id specified to update their mac
addr at plc. before, it also had to have a 3.x cd.
Aaron Klingaman [Thu, 28 Jul 2005 22:59:20 +0000 (22:59 +0000)]
bump version of bootmanager for mac addr/ifconfig changes
Aaron Klingaman [Thu, 28 Jul 2005 22:58:35 +0000 (22:58 +0000)]
- after reading network configuration file, fill in mac address if missing
- when calling getnodeid.php for nodes without node_id value, post mac addr
not ifconfig
Aaron Klingaman [Thu, 28 Jul 2005 22:57:20 +0000 (22:57 +0000)]
from
http://forums.devshed.com/archive/t-51149/Ethernet-card-address-Through-Python-or-C
add a function that will return the mac address for a given network device
Aaron Klingaman [Thu, 28 Jul 2005 22:56:29 +0000 (22:56 +0000)]
don't post ifconfig output when uploading logs, no longer necessary
Aaron Klingaman [Thu, 21 Jul 2005 23:16:27 +0000 (23:16 +0000)]
for node_id/plconf change, need to get variable PLCONF_DIR too
Aaron Klingaman [Thu, 21 Jul 2005 21:50:55 +0000 (21:50 +0000)]
now that PlanetLabConf uses node_id, update the bootmanager code that calls
PlanetLabConf and write out /etc/planetlab/node_id.
Aaron Klingaman [Wed, 20 Jul 2005 16:53:10 +0000 (16:53 +0000)]
minor revision bump for node_id file change
Aaron Klingaman [Wed, 20 Jul 2005 16:51:38 +0000 (16:51 +0000)]
make BootManager write out /etc/planetlab/node_id
Mark Huang [Wed, 22 Jun 2005 22:28:21 +0000 (22:28 +0000)]
- bump version to 3.1.1 to reflect USB fixes: support booting from USB,
shutdown USB HCDs before chain booting, except when booting from USB
Mark Huang [Wed, 22 Jun 2005 22:21:53 +0000 (22:21 +0000)]
- fix some log messages
- get usage count for USB (currently, just the usage count for
usb_storage); if in use, don't try to unload the HCD drivers. This is
an ugly workaround; the real solution is to fix all the HCD drivers to
shutdown properly in the kexec path, and not unload them manually
ever. A node will have problems if it both boots from USB, and its HCD
shares an IRQ with another device. So far, it's been either one or the
other.
Mark Huang [Tue, 21 Jun 2005 21:24:09 +0000 (21:24 +0000)]
- when unmounting floppy or usb keys, unmount the mount point, not the
device, which (in the case of the usb key being used for both the boot
device and the configuration device) may be already mounted.
- loop through all partitions of removable devices looking for
configuration files.
Mark Huang [Thu, 16 Jun 2005 19:34:50 +0000 (19:34 +0000)]
- PL6577: at least on the ufl.edu nodes, the USB host drivers need to be
unloaded before chain booting, since they share an interrupt with the
e1000 cards, and don't cleanly shutdown() the hardware when kexec is
run.
Aaron Klingaman [Wed, 8 Jun 2005 17:54:59 +0000 (17:54 +0000)]
back out lvm extent size change␓
Aaron Klingaman [Tue, 7 Jun 2005 21:03:59 +0000 (21:03 +0000)]
reduce lvm physical extent size to 16M from 32M. This was the cause of the
install problems on planetlab1.mini.pw.edu.pl which had two identical
western digital 120gb drives. This also limits the maximum size of the
vserver logical volume to 1TB, it was about 2TB before this change.
Aaron Klingaman [Mon, 6 Jun 2005 14:51:47 +0000 (14:51 +0000)]
rebuild doc, previous one used wrong utils and didn't include graphics
Aaron Klingaman [Thu, 2 Jun 2005 22:06:17 +0000 (22:06 +0000)]
send a nice message to tech contacts if the node hostname resolution is
incorrect or doesn't work
Aaron Klingaman [Thu, 2 Jun 2005 20:00:41 +0000 (20:00 +0000)]
don't unnecessarily shutdown/startup lvm between successful install and
chain boot
Aaron Klingaman [Wed, 1 Jun 2005 19:54:57 +0000 (19:54 +0000)]
better handle the kexec failure cases by notifying the user of the exact
problem, and canceling further boot processes
Aaron Klingaman [Tue, 31 May 2005 17:30:57 +0000 (17:30 +0000)]
CACERT_PATH still in /mnt/cdrom, will be moved later to /etc/planetlab
Aaron Klingaman [Tue, 31 May 2005 17:14:31 +0000 (17:14 +0000)]
update boot manager tech docs post implementation/deployment
Aaron Klingaman [Tue, 31 May 2005 16:44:03 +0000 (16:44 +0000)]
move support files to /boot
Aaron Klingaman [Tue, 31 May 2005 16:43:30 +0000 (16:43 +0000)]
various source file cleanup, unused configuration vars removed, move
support files from /alpina[-v3] to a common /boot
Aaron Klingaman [Sat, 28 May 2005 01:20:56 +0000 (01:20 +0000)]
so some sata devices show up in lspci under the ide device class,
and some show up in the scsi device class. for the pitt nodes, its the
former. this fix works by assuming that if an ide pci device matches
a loadable module, then its sata, because the ide modules are built in.
it will then show up in /etc/modprobe.conf under a scsi_hostadapter
alias.
Aaron Klingaman [Thu, 26 May 2005 19:19:49 +0000 (19:19 +0000)]
be extra clear about what happens when a node that doesn't have node_id
on the floppy is attempted to be booted, and it doesn't yet exist in PLC.
this is the case for when users try to add nodes using the old procedures
without reading the guides (bound to happen)
Aaron Klingaman [Thu, 26 May 2005 17:35:49 +0000 (17:35 +0000)]
update docs about search path for network configuration floppy
Aaron Klingaman [Thu, 26 May 2005 17:27:26 +0000 (17:27 +0000)]
the beta 3.0 cds didn't have uudecode, so its put here. like other files
in this directory, this ends up on the boot server for download by the
bootmanager as needed
Aaron Klingaman [Thu, 26 May 2005 17:25:03 +0000 (17:25 +0000)]
move bootmanager (was alpina) support files into this repository
Aaron Klingaman [Thu, 26 May 2005 17:11:48 +0000 (17:11 +0000)]
check in all bootmanager sources
Aaron Klingaman [Thu, 26 May 2005 17:10:19 +0000 (17:10 +0000)]
this will have to suffice until a proper make/build system is created.
this compresses the bootmanager sources, base64 encodes them, and creates
a single bash script that does the reverse of that, and executes the
bootmanager
Aaron Klingaman [Tue, 15 Mar 2005 22:27:56 +0000 (22:27 +0000)]
shell commands don't need to be in $(...)
Aaron Klingaman [Tue, 15 Mar 2005 22:02:48 +0000 (22:02 +0000)]
add the built file too, since it won't be changed much and requires nasm to build
Aaron Klingaman [Tue, 15 Mar 2005 21:57:06 +0000 (21:57 +0000)]
a simple boot loader that will be copied to all disks that outputs a message
telling the user that the machine requires a boot cd
Aaron Klingaman [Tue, 15 Mar 2005 14:29:03 +0000 (14:29 +0000)]
added network configuration file section; built pdf
Aaron Klingaman [Mon, 14 Mar 2005 22:57:19 +0000 (22:57 +0000)]
add reference to pdn, describe boot states, expand upon authentication
Aaron Klingaman [Mon, 14 Mar 2005 15:15:24 +0000 (15:15 +0000)]
should have been exported in something other that black and white, this is actually readable
Aaron Klingaman [Mon, 14 Mar 2005 15:08:01 +0000 (15:08 +0000)]
always send hardware to plc when booting, remove unneccasry extra check for ip address
Aaron Klingaman [Thu, 10 Mar 2005 20:02:49 +0000 (20:02 +0000)]
*** empty log message ***
Aaron Klingaman [Thu, 10 Mar 2005 19:58:59 +0000 (19:58 +0000)]
tech docs for boot manager
Aaron Klingaman [Thu, 10 Mar 2005 19:57:13 +0000 (19:57 +0000)]
moved
Aaron Klingaman [Mon, 7 Mar 2005 18:10:14 +0000 (18:10 +0000)]
added built pdf document
Aaron Klingaman [Fri, 21 Jan 2005 15:49:31 +0000 (15:49 +0000)]
switching to apache fop to build pdfs, much better output, so
no longer need the build script
Aaron Klingaman [Thu, 20 Jan 2005 22:47:55 +0000 (22:47 +0000)]
used to build pdf output for pdn
Aaron Klingaman [Thu, 20 Jan 2005 19:27:01 +0000 (19:27 +0000)]
title change
Aaron Klingaman [Fri, 14 Jan 2005 19:29:33 +0000 (19:29 +0000)]
oops, should have named the document more appropriately