i was working on a brand new 2013 laptop yesterday. i spent hours (no exaggeration) on this thing. and i had my first moment where i felt old and out of touch with the current technology. most notably all the UEFI stuff. it really made me miss the antiquated BIOS.

anyways, i wanted to load windows 8.1 on the laptop. mainly because i needed to repartition the drive to make room for linux. it seems very hypocritical that i would have anything to do with windows 8 since i loathe it, but i wanted a new OS for new hardware. and i’ve determined that i can live with windows 8 w/ classic shell installed and booting directly to the desktop. when you do this it’s easy to forget that the metro UI even exists. i take a similar approach with ubuntu by using gnome classic instead of the gnome3 UI or unity.

step #1
get into UEFI settings. this took me an eternity to figure out. i hit F2 on boot and every other key i could think of. i finally determined that if UEFI Secure Boot is enabled then you can only get to the UEFI settings from the installed OS. in windows 8 this involves holding down the Shift key while choosing to restart the computer. there’s also another method that involves digging through menus. once i got into the UEFI settings i disabled Secure Boot. i could then hit F2 to get into the settings without involving windows. i also noticed i had to hit F2 almost immediately when powering on the laptop.

step #2
figure out how to boot off install media. i was sitting around burning dvds for install media (and having some troubles with this as well) when i suddenly looked at the new laptop and realized it had no optical drive. i just took it for granted that everything came with optical drives. crap i thought…what am i going to do now? i looked around for USB optical drives and came up with nothing. then i started googling how to boot from USB media.

thankfully, for windows it was incredibly easy. download the windows 7 usb/dvd download tool (http://www.microsoftstore.com/store/msusa/html/pbPage.Help_Win7_usbdvd_dwnTool). even though it’s named “windows 7…” it works fine with a windows 8.1 iso. the ubuntu equivalent is an ubuntu packaged named “usb-creator” which comes in qt and gtk versions. the ubuntu utility is also equally simple to use.

at this point i had my install media problem taken care of, but then i couldn’t figure out how to actually boot from the USB devices. i spent a long time researching this as well. in the UEFI settings there’s a UEFI Boot List Option and Add Boot Option. i found the Add Boot Option to be very cryptic and couldn’t find any useful info online for adding USB devices. i eventually gave in and changed the UEFI Boot List Option to Legacy. then bam! i suddenly had options for booting from CD/DVD, USB device and so on. what i find puzzling and frustrating is why are these basic tasks more difficult to accomplish in UEFI than they were in BIOS. i was able to alter the boot order accordingly and use F12 for a one time boot menu after this.

UEFI setting changes:
Secure Boot = Disabled
Boot List Option = Legacy
Intel Rapid Start Technology (Fast Boot) = Disabled

i ended up disabling Fast Boot because it was a recommendation on many linux forums, not because i was experiencing any actual issues with it.

so what do you lose for changing from UEFI boot mode to Legacy? essentially, you lose the ability to boot from a 2TB+ partition (which wasn’t even applicable in my case).

step #3
apparently, changing from a GPT disk to an MBR disk in the method that i did leaves behind stray GPT data that confuses some partitioning utilities like gparted. this reared its ugly head when i was in the middle of the ubuntu install (during the disk partitioning step). the ubuntu install saw the disk fine, but did not display the existing NTFS volume. i quit the install and did some research. i found that someone made a utility specifically for fixing this named FixParts (http://www.rodsbooks.com/fixparts/). i ran the windows version of the utility like so:
fixparts 0: (from administrator command prompt)

upon running fixparts i was initially presented with the message “NOTICE: GPT signatures detected on the disk…”. i entered yes to delete the GPT signatures and then quit the utility. that was all that needed to be done in my case.

step #4
install nvidia driver in ubuntu. the ubuntu install defaulted to the open source nouveau driver which was running fine. nonetheless, i preferred to have the full nvidia driver installed. for so long ubuntu has spoiled me with the simplicity of installing this driver (just install the package nvidia-current). this time however i received a black screen when starting my gnome sessions after installing the driver. i tried the usual tricks of reconfiguring X and so on. i didn’t have any luck and finally gave in and started researching online. it turns out if you have a modern laptop with the intel hd graphics + nvidia card combo, then you need to install the following packages:
sudo apt-get install bumblebee bumblebee-nvidia (source: http://askubuntu.com/questions/372594/problem-with-installing-nvidia-display-drivers-on-ubuntu-13-10)

i was in good shape after installing bumblebee.

and lastly some tweaks i applied to…
ubuntu
installed kompozer from the packages here: http://linuxg.net/how-to-install-kompozer-on-ubuntu-13-04-12-10-12-04/

move window buttons to righthand side…this used to be done with gconf-editor but not in ubuntu 13.10. it’s done like so:
gsettings set org.gnome.desktop.wm.preferences button-layout ‘:minimize,maximize,close’
source: http://askubuntu.com/questions/370038/ubuntu-13-10-move-the-window-buttons-to-the-right

there seems to be some issues with newer intel centrino wireless-n cards (in my case the 7260) and the linux driver iwlwifi. it’s an annoying issue where your wireless connection suddenly drops to 1mbps (but actually it’s a lot less than this). file transfers/downloads are nearly impossible when this occurs.

sudo lshw -C network
*-network
description: Wireless interface
product: Wireless 7260
vendor: Intel Corporation
physical id: 0
bus info: pci@0000:02:00.0
logical name: wlan0
version: 73
serial: fc:f8:ae:8b:67:f8
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress bus_master cap_list ethernet
physical wireless
configuration: broadcast=yes driver=iwlwifi
driverversion=3.11.0-15-generic firmware=22.0.7.0 ip=192.168.1.30
latency=0 link=yes multicast=yes wireless=IEEE 802.11bgn

the issue was resolved for ubuntu with the 2/6/2014 update named “Firmware for Linux kernel drivers”, which updates firmware version 22.0.7.0 to 22.1.7.0. prior to that update’s release, i was using the temporary workaround of running:
sudo rmmod iwlmvm
sudo rmmod iwlwifi
sudo modprobe iwlwifi 11n_disable=1

this essentially disables wireless-n capabilities and you are capped at wireless-g speeds. the connection was stable after applying this.

sources:
http://syntaxionist.rogerhub.com/intel-centrino-wireless-n-2200-ubuntu-1mbps-workaround.html
http://askubuntu.com/questions/119578/how-to-fix-slow-wireless-on-machines-with-intel-wireless-cards

windows 8.1
disable smartscreen filter in action center
info: http://www.howtogeek.com/75356/how-to-turn-off-or-disable-the-smartscreen-filter-in-windows-8/

also very cool that you can natively mount .iso files in windows 8.1 (maybe 8 too?) now.

and i am saddened that you apparently can’t run 16-bit apps under windows 8.1 64-bit…

very recently…a dell c1765 mfp was deployed and installed as a printer on a papercut secondary print server. jobs printed to this print queue were listed as 0 pages and thus not being charged the appropriate cost per page. the issue was isolated to this particular printer. unfortunately, there was no alternative driver to test out because only the initial release was available.

papercut release 14.0.25901 resolves the issue with this printer.

update 1/28/2014:
correction, the papercut 14.x release resolved the 0 pages/0 cost problem, but after applying it the dell printer queued up jobs and simply deleted them without printing. i upgraded the printer firmware from 1.02.05 to 1.22.00 and also used a newer windows 7 driver listed as version 1.0.2.3 (dell version A02)…then everything worked correctly.

i encountered an oddity today for the first time. most of my experience with windows 2003 is based upon windows 2003 R2…so that’s probably why i’ve never seen this before. in the non-R2 release when you try to RDP to the local console user’s session it initiates a separate session. doesn’t seem like a big deal right? well it is because you essentially get a blank screen and no desktop.

i tried messing around with various RDP client settings to no avail. eventually found out that in this windows server release, you need to specify the computer/IP address in the RDP client and append “/admin” at the end. apparently, in even older RDP client releases this switch was “/console”.

source:
http://weblogs.asp.net/jeffwids/archive/2005/12/14/remote-desktop-to-console-session-on-windows-server-2003.aspx

for additional security, SMB signing can be enabled on windows servers. however, according to microsoft and other sources, enabling this feature can degrade file transfer performance up to 15%. so it may not be a good idea to enable it on heavily used file servers…but might as well enable it on servers that do not experience a lot of SMB traffic.

here’s where to enable it:
Local Security Policy->Security Settings->Local Policies->Security Options
Microsoft network server: Digitally sign communications (always)
or
Microsoft network server: Digitally sign communications (if client agrees)

source:
http://technet.microsoft.com/en-us/library/cc731957.aspx

charles recently brought this issue to my attention. we’ve had more than a handful of outlook 2013 users complain about calendaring issues and being unable to connect to exchange servers (and so on). turns out the problem is caused by having outlook 2013 and certain MS updates installed. microsoft released a fix for the issue on dec 13th, 2013: http://support.microsoft.com/kb/2837618/en-us

sources:
http://www.infoworld.com/t/microsoft-windows/botched-outlook-2013-patches-kb-2837618-and-kb-2837643-break-out-of-office-reply-freebusy-and-more-232055
http://helpdesk.missouristate.edu/news/two-microsoft-updates-cause-problems-with-outlook-2013

i noticed this issue nearly a month ago. after a host would be running for a couple of days and joined to vcenter the resources pane on the summary tab would look “blank” (as in cpu usage: 0mhz, memory usage: 0mb). the first time i saw it i rebooted the host and the issue went away.

later on, i saw the issue return to the same host and then also spotted it on other hosts. it was obvious this was a bug at the time (since there were powered on vms on these hosts). however, i wasn’t immediately that concerned with it. a reboot would temporarily resolve it.

this bug started to annoy me when i actually needed to vmotion vms around as i was basically driving blind without resource stats. still, i didn’t experience real concern until i noticed that DRS was believing these false stats and completely evacuating hosts that were affected by the bug. i couldn’t have perfectly healthy hosts running zero vms. at this point i decided to look into it and found it listed in a vmware kb article. the issue is resolved by upgrading to vcenter 5.1u1a or later.

source:
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2038686

if you have sql logins that are based on AD users and the AD username changes at some point, these changes are not updated automatically in ms sql. i’ve read various posts that claim that restarting the sql service or the server entirely will cause the sql login info to update. i have yet to see this actually happen.

when there’s this mismatch between sql login names and AD usernames, functionality doesn’t suffer, but the sql error logs will grow to massive proportions. the Application log in windows event viewer will likewise be flooded with sql errors.

since ms sql 2005, there has been a sql command to resolve this issue:
ALTER LOGIN [domain\user] WITH NAME = [domain\user];

i have tested this out and it works great.

source:
http://social.msdn.microsoft.com/Forums/sqlserver/en-US/a41f785a-fe6e-4c87-b90f-e3d8dd4a4e8a/renaming-domain-user-accounts-and-sql

haven’t been on 5.1 too long and i’m already looking into moving to 5.5. i’m motivated by the fact that there are some critical shortcomings (some new, some longstanding) that are addressed in 5.5. in addition, 5.1 seems to be a somewhat buggy release.

i’ve been reading through various kb articles and the install guide this morning and it appears that this upgrade should be smooth sailing (in contrast to the 4.1 to 5.1 upgrade). i’m feeling fairly confident that this upgrade won’t have any problems (probably jinxing myself).

since i have all the components installed on a single server, i am going to try out the “simple install” method.

the same vcenter 5.x license key will work with 5.5 (previously i thought i had read otherwise). excellent license key matrix here: http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&docType=kc&docTypeID=DT_KB_1_1&externalId=2059926

other than that, the only item i’ve found noteworthy is that the SSO admin account will be automatically changed from admin@system-domain to administrator@vsphere.local.

source:
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2053130

update 1/24/2014:
i applied the vcenter 5.5 update around 2 weeks ago. i used the simple install method and it was the easiest vcenter upgrade i’ve ever been through. i highly recommend using this method if you can. and really there shouldn’t be anyone that can’t as it’s recommended best practice to co-locate all the vcenter roles onto the same server (just recall that you need a decent amount of RAM nowadays for vcenter). the only issue i ran into was with VUM and that was because i had been struggling with VUM since vcenter 5.1. the VUM service would cease functioning but refuse to be stopped or restarted. only a reboot would resolve it. so during the VUM upgrade since the service couldn’t properly stop the upgrade of course failed. here’s the error that given – http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2059567. to resolve i rebooted the vcenter server and tried the VUM upgrade again and it was successful. also my VUM issues have disappeared since the upgrade to 5.5.

and…recently confirmed that after you upgrade to vcenter 5.5’s SSO module (which has been dubbed SSO 2.0) you can remove the old SSO database (which should be named “RSA”). my suspicions have been confirmed as well that windows based vcenter will eventually go away. the VCSA is the future and will most likely be feature complete for the vcenter 6.0 release. also the windows based vsphere client (sometimes called the C# client) support goes away after vsphere/vcenter 5.5 (this shouldn’t be a surprise to anyone at this point). VUM will be going away and will be replaced by similar functionality in vcenter (most likely debuted in vcenter 6.0). and finally vsphere/vcenter 6.0 will probably cut off support for virtual hardware level 4 (this hasn’t been confirmed yet but it’s a strong rumor).

check SSO functionality go here:
https://vcenter.mydomain.com:7444/lookupservice/sdk

this will be my last post concerning VDR. recently, experienced an issue where i could no longer connect to the appliance via the VDR plugin. i tried a couple of things…restarting the VDR service on the appliance, restarting the appliance, and even uninstalling/reinstalling the VDR plugin. none of the above worked.

eventually found an article that suggested deleting the following files:
c:\windows\system32\libeay32.dll
c:\windows\system32\ssleay32.dll

i was initially skeptical about this approach. i didn’t even have the ssleay32.dll file present on my system. anyways, i ended up trying it by deleting libeay32.dll and it worked out fine.

source:
http://h10025.www1.hp.com/ewfrf/wc/document?cc=us&lc=en&docname=c03366025