Saturday, November 27, 2010
If you're building your own WiaB and get a lot of "cannot connect" errors, take a look at what interface your server is listening on (i.e., use netstat). Even though I'd properly configured the forward and reverse lookups in DNS, I'd forgotten that Linux is usually configured to check /etc/hosts first. Since I'd left the `hostname` code in the config file (where "hostname" is surrounded by backticks), it was actually defaulting to "localhost" vice the proper IP address for the server. (Note: you probably won't receive this error unless you're running WiaB and your XMPP server on different systems.)
Experimenting with federation took a bit longer as it requires configuration of DNS, /etc/hosts, and SSL certificates. Nothing too difficult though. End result is that I now have a WiaB instance tied to the Openfire server. Notes in the wiki shortly.
Friday, November 12, 2010
I didn't like the plugin because it sends you the entire email, vice sender/subject, via Jabber. It's quite messy when viewed in Xchat via Bitlbee.
For now, there's no rush for me to experiment heavily with these. I have to take the main server down to play with these and current house guests get cranky if the "Internet is down" during the evening hours. It probably won't become a crisis until VMware removes para-virtualization from ESXi 4.x (supposedly soon).
Xen XCP looks like it's ready for someone to write a new interface. The current one involves Java and I learned to dislike Java control GUIs back when Gauntlet 4.0 was cutting edge. Nothing's simple and there's too much stuff missing. Haven't had a chance to install the management software for Citrix, so I can't yet comment on that.
I'll keep notes in the wiki as I go along.
Tuesday, November 9, 2010
Just before I generated the preceding post, I'd started a very specific Google search. As soon as I hit "Publish Post", I went back to the Google search page and hit "Next". There, at the bottom of the next page, was the post that I'd just generated. If Google gets any faster, I'll have search first so's I can read what I'm about to write. (heh)
Saturday, November 6, 2010
By default, the text-to-speech script that comes with the Liquidsoap audio scripting language is intentionally broke (uses "/bin/false" as the TTS tool!). The fix is easy:
- Edit /usr/local/lib/liquidsoap/0.9.3/liquidtts
- In the first line that starts with "echo", change the word "false" to "text2wave"
The next to the last line shouldn't need the word "false" changed to "text2wave". It's there in case the first part (before the first "return") fails.
And that's it. If you add say_metadata to your liq script, you should hear a familiar crappy synthesized voice, telling you what the song was.
If you want to use swift (i.e., Cepstral), the following seems to work well:
# This script is called from liquidsoap for generating a file
# for "say:voice/text" URIs.
# Usage: liquidtts text output_file voice
swift -o $2.wav " that was $1 " && /usr/bin/sox $2.wav -t wav -c 2 -r 44100 $2 2> /dev/null > /dev/null
false $2 2> /dev/null > /dev/null
Please note that the first line of the script starts with "swift" and ends with "/dev/null > /dev/null". Blogger likes to wrap code and cause problems.
I added in the "prosody" tags to slow Allison's speaking a bit. The default rate is set for "New Yorker". I've spent too much time in the South to be comfortable with it for any length of time. If you want to tweak the voice further, Cepstral has some common usage examples.
I'm still trying to figure out how to get pass/accept events in and out for speech and for posting events to Twitter or IRC. The former probably means having to pull from a secondary stream. For the later, I'm pestering people (who said they had it working) for hints.
More reading required on my part, I guess.
Friday, November 5, 2010
On the candidate list is Citrix's free XenServer and Xen's Xen Cloud Platform. Having previously run the build-your-own-Xen-kernel gauntlet and being reluctant to go through it again, this is probably going to be the entire range of choices. If anyone knows of other choices, other than KVM, VirtualBox, and VMware's lesser offerings, please yell out.
Wednesday, November 3, 2010
The good news is that ESXi has minimized a large chunk of my usual rework (e.g., rebuilding the main system every 3-4 months and annoying Sparks to no end). This past weekend, I reached the point where I was attempting to install the new version of the LiquidSoap scripting language (see previous post) and then writing fixes for a Perl jukebox on top of that.
The bad part is that I've come to the realization that I'm so out-of-practice that I can't "see around the corner" while writing code. Embarrassingly, I had to drop back to methods learned a quarter century ago: composing flow charts, single-stepping through same, and tracking state on paper (bleh!). It's almost as bad as having to document someone else's code.
The good news is that I've taken good notes over the years and much of the coding can be done cookbook style (+1 for modules!). Hopefully it won't take too long to "blow the carbon out of the cylinders" and get back up to speed. I've got a couple weeks-worth of documentation to write and (hopefully) will have time to get back to coding in Perl and C.
Monday, November 1, 2010
The short version is that I dropped back to v0.9.3 and it compiled without an issue. I'm now tweaking the in-house jukebox (a web front end to a liq script) so that it runs, spread across multiple VMs. Next up is to incorporate text-to-speech and see if it will work with Cepstral in a Festival wrapper. If that works well, things will probably get weird very quickly.
Tuesday, October 26, 2010
Sunday, October 24, 2010
Then again, it could be my machine. My serial port is actually located on a PCI card and not the motherboard. More than likely, I'll need to resort to adding drivers to ESXi again (not something that I'm looking forward to).
I'll keep poking at it and let y'all know if I get it running. I may have to resort to a eth-to-serial adapter.
Saturday, October 23, 2010
account add twitter yourname yourpass
account 2 set base_url http://yourserver/api
account 2 set mode chat
account 2 set oauth false
account 2 on
The problem is that, if fancy URLs aren't enabled, the API line is incorrect. The fix is use of a slightly different syntax:
account 2 set base_url http://yourserver/index.php/api
Sunday, October 17, 2010
Adding the old content to the new wiki is in progress (slow going due to time constraints). Currently, about half of the Asterisk content is posted. If there's to-be-restored content that you want prioritized, give a yell.
The movement is a bit slow due to school, re-certification requirements, and rebuilding a few VMs. (Note to self: beat self up for not testing that last piece of software on a scratch VM and for not recently employing the snapshot or backup functions.)
For Sparks, the rebuild is almost complete. We need to look at resurrecting the VPN.
Tuesday, October 12, 2010
The URL for the new wiki is: http://www.757.org/~joat/wiki/.
Sunday, October 10, 2010
Short version of the story is bip 0.8.6 and bitlbee 1.3dev are now installed, both compiled from source code and both playing nicely.
However, I did spend most of this morning trying to get XChat to autoconnect to bip (and bitlbee/Freenode/OFTC/757 beyond that). Finally figured it out: you shouldn't add the "quote pass name:passwd:connection" string to the autoconnect command field. Instead, just add "name:passwd:connection" to the Server Password field. If you're editing the servlist_.conf file by hand, your server will have a "P" line that looks like:
I'd been messing with the "C" line and not getting anywhere. It wasn't until I was wading through the XChat docs that I realized that this might be treated as a password string. Tried it and it worked!
Why is it that the simple things take all day?
Sunday, September 19, 2010
account add twitter YOURNICK YOURPASSWD
Next, figure out the account ID by running:
Assuming that it's the first account (ID == 0), turn on the account via:
account 0 on
This will spit up a message similar to:
twitter - Logging in: Requesting OAuth request token
*twitter_packetgeek* To finish OAuth authentication, please visit
ao&oauth_callback_confirmed=true and respond with the resu
lting PIN code.
Please note that everything after *twitter_packetgeek* in the above is all on one line.
You'll want to make note of the handle located between the *'s in the second line (in my case, twitter_packetgeek). Copy and paste the URL into your browser (you may need to log into Twitter) and click allow. This will give you a number to paste into the irc client.
Here's the undocumented part for the irc client: you need to MSG the twitter_packetgeek account with the token. Do this via:
/msg twitter_packetgeek XXXXXXX
where "XXXXXXX" is the number generated by Twitter's Oauth.
Wednesday, September 15, 2010
In any case, I hope to have it online this weekend. Same server, slightly different URL.
Tuesday, September 14, 2010
Jim, a classmate, learned that you could add drives to FreeNAS without shutting the VM down. We're still not sure if this is "good practice" (or even a feature) but it's interesting none-the-less. For some reason, his new drives show up as SCSI.
I'm still finding stuff in the menu to play with. SNMP, LDAP, and WEBDAV have piqued my interest. I'm wondering how well FreeNAS would work as back-end storage for Alfresco and/or KnowledgeTree. Nagios looks like it'd make a good monitoring tool.
Configuration notes are coming. The wiki migration is 99% done (I have a few dozen pages to transcode). I'll start adding new notes once the new wiki is stood up.
Sunday, September 12, 2010
- my Bloglines subscriptions are dropping at a fast rate (down to about 170 now)
- my Delicious bookmarks have jumped up
- my Google Reader subscriptions have jumped up
While moving subscriptions for active feeds, I noticed a new follower in Google Reader (Tims Club). I can't describe (yet) why I don't have a good feeling about this. My initial impression is this is something akin to referrer spam. I'll probably have more to write about this later.
Update: I managed to discover the "select range" feature and the trash can. The last 170+ subscriptions went away in one swell foop. The feature is in the left-hand menu. Click on "Edit" and scroll to the bottom. Read the text with the yellow background.
Monday, September 6, 2010
In any case, I can now start/stop the FreeNAS server via the ESXi console. Notes in the wiki shortly (I have a backlog, don't I?).
Update: I've received a couple requests for the packages (one from v1ncen7 below). The howto is here. The link for a tarball of the generated packages is at the bottom of that page and here. The usual disclaimers apply (i.e., your mileage will vary).
Friday, September 3, 2010
Initially, logic led me to believe that "vm-cmd vmsvc/power.off 32" would be the obvious shutdown command. It did turn the VM off. Problem was that it was extremely fast. Too fast. A little research showed that "vm-cmd vmsvc/power.off 32" was the equivalent of holding the power button in on the machine (i.e., no shutdown sequence).
A bit more research shows that I should install vmware-tools and use "vm-cmd vmsvc/power.shutdown 32". It's still quick, but I've logged into the machine to watch what happens and it is going through its shutdown sequence.
So, I now have the background tools to be able to start/stop a number of development systems via a web interface. I only need to do a little coding to get it to work.
Note: "32" in the above examples is the vmid for one of my VMs. Yours will probably vary. You can list yours by running "vm-cmd vmsvc/getallvms".
Update: If you're running 64-bit Ubuntu VMs, you'll likely need to install open-vm-tools and open-vm-dkms before power.shutdown will work. Apparently vmware-tools doesn't work for 64-bit systems. Can anyone very that this is accurate?
Wednesday, September 1, 2010
I also want to experiment with other VMs (Astaro, Koha, Plutex, etc.) but know that there aren't enough resources to leave more than one of them running without affecting the other two. End result is that I need a web interface to manage which ones are running. It'll involve a bit of obfuscation, due to VMware's restriction on remote scripting for the free versions, but it shouldn't be that difficult.
I'll post my notes in the wiki when they're ready.
Saturday, August 28, 2010
A direct install of ESXi 4.1.0 on my old HP kept erroring out (couldn't access the SATA storage). An attempt to build a custom install disk, reusing the oem.tgz file, didn't work. Building yet another oem.tgz file was obviously going to be the method-of-last-resort because it's been a year since the last time and that involved two days of trial and error.
So, finally, I broke down and started reading the forums. Surprise! VMware has an upgrade bundle for moving from 4.0 to 4.1. I downloaded the zip file, put the test system in maintenance mode, and ran the esxupdate tool on the command line. It installed without a hitch.
Same for the house server. It's now running 4.1.0. I was anxious that it'd replace all of the drivers (the house server was built using last year's custom install disk) but it appears that it didn't touch any of those. The hypervisor can still see both datastores and can still access the network.
Note to self: If you have to install this again, remember that the system needs to be put in maintenance mode BEFORE attempting to run esxupdate.
Note to Dave: I think that the 4.0 custom disk is the same one that we used to build your system. This should be a quick upgrade to your system, too.
Thursday, August 26, 2010
It is painful to stop using MythTV but it frees up the display that I was using with it. This blow is softened by my wife having acquired a DVR to act as a replacement.
Another reason is that the third class on virtualization (actually storage for virtualization) is about to start and I want a system at home so that I can keep up with the class.
Things to do first:
- back up everything!
- figure out what a 4.0 to 4.1 upgrade will break
- install the new drives (upgrade to 4 TB)
- break it gently to Sparks that I'm going through yet another system upgrade (he gets annoyed that I do this every two months or so), though it'll minimize future down-times as VMs can be built before others are taken offline
- determine if 4.1 can be modified to include needed drivers (the 4.0 install did) (my NICs are older)
- get FreeNAS installed
The tricky part is going to be efficiency. While I could build a single server VM which runs everything, it makes backups difficult. I think it'll be easier to manage if I can create a number of very small servers, dedicated to specific tasks (Icecast, Squid, etc.).
Ideally, they should be started and stopped on demand. I need to do a bit more research on this. I think the problem is going to be that VMware has limited the command line functions to "read only" for the free versions of ESXi. I'll keep you posted.
Wednesday, August 18, 2010
I'll post my notes in the wiki shortly. The one thing that I'd add to AW's work concerns the installation of phpLDAPAdmin. He left out that, if you change the container, you'll need to edit /etc/phpldapadmin/config.php. Basically, look for the uncommented lines which contain "dc=example,dc=com" and change the values to whatever you used in building your directory. If you don't, phpLDAPAdmin will always try to access example.com and will error out, even if you change the domain on the login screen.
Tuesday, August 17, 2010
The issue turned out to be the firewall (the ufw service). You'll need to allow port 3000 (or whatever port you used for Ntop if you didn't use the default). Tell ufw to allow TCP packets on port 3000 via:
sudo ufw allow 3000/tcp
The above, and other ufw commands, can be referenced on the Ubuntu Community ufw page.
Saturday, August 14, 2010
1) Install ntop via "sudo apt-get install ntop"
2) Set the admin password via "sudo ntop --set-admin-password"
3) Start the ntop service via "service ntop start"
4) Point a browser at http://localhost:3000
Monday, August 9, 2010
Getting an account was easy, as was accessing the main page. However, with the default settings in Chrome, I cannot access any of the tools (Word, Excel, etc.). The browser keeps complaning that the certificate for the web site has been revoked. In looking at the certificate, the CRL is hosted on a Microsoft server.
Has anyone else noticed this? Is it a new thing?
Oh, and yes, I can access the tools if I turn off CRL checking (not a good idea for the security concerned).
Sunday, August 8, 2010
Friday, August 6, 2010
At one point, this blog was associated with malware analysis (the posts are in the archive). I've been too busy (for years) to touch much on the topic. I wonder if I'm a victim of my own nosey-ness, anywhere else...
Wednesday, August 4, 2010
Monday, August 2, 2010
- DNS entries (forward and reverse) are a must. Host table lookups weren't working.
- The OPER password has a not-safe-for-production work-around (ask me sometime).
Saturday, July 31, 2010
The good news is that it didn't take too long to also migrate the graphics from 757 to Blogger. The bad news is that, over time, four pictures were somehow lost. (Don't know how that happened.) Also, the format of some of the older posts is a bit borked (I'll work on these over time).
The previous attempt, using the Google Blog Converters AppEngine, didn't go well. The dates for the posts were all the same.
This attempt was moderately sedate, though a bit involved. I imported the Blosxom content into Wordpress and then used the online Wordpress-to-Blogger tool to export from Wordpress and import to Blogger. The latter tool has a limitation in that it doesn't export files bigger than 1MB. This was worked around by limiting the Wordpress output to individual years. It's a little bit of extra work but it took longer to write this post than it did to export all of the posts.
In any case, the old blog is down. Enjoy this one!
Tuesday, July 27, 2010
In any case, I was able to geek a bit, in between afternoon thunderstorms. It took a bit of work but I did get Asterisk 1.8.0beta1 compiled. For those considering it, be sure that you know how to track down the odd development library and how to compile third party code. There's a number of external packages that you need to find/compile. I'm researching the iCal, IPv6, and Jabber interfaces. I'll add notes to the wiki as I learn about them.
The power supply issue is going to be painful. It looks like we may have to spend almost 50% of what the box cost us to get an UPS with "adequate" power conditioning. Faugh on you, HP!
The Ubuntu install on the Acer Revo went without a hitch. I added an external SATA drive, created the boot stick with usb-creator, adjusted the boot selection in BIOS, and installed 10.4 in about the usual time, all without affecting the native XP install. Audio, video, and network all worked nicely without having needing repair. I now have the little box connected to a KVM with the main box and have the vSphere client installed on XP.
I didn't get much work in on reviewing Jabber bots. I was too distracted in reading the docs for Asterisk 1.8
I have a backlog of stuff to post to the wiki too. Maybe this weekend?
Wednesday, July 21, 2010
- work on friend's ESXi system, get various sample systems running and documented
- tweak on own ESXi box
- install Ubuntu 10.4 ISO on thumb drive and use it to dual boot on an Acer Revo (no external drives)(adding .5TB SATA though)
- attempt to "adapt" Bitlbee code for alternate purposes
- review a number of Jabber bots
- find a usable UPS or line conditioner for an insanely line-noise sensitive HP Proliant 350 G6 (dang thing quits if there's a lightning strike two counties over)
- wade through the still-growing backlog of photographs of family and friends
And that's just the geek stuff. I still have to mow the lawn and do battle with the hedges (they always win).
Monday, July 19, 2010
In any case, the 4.1 install didn't work. It would boot but would fail early in the install, claiming that it couldn't see my CDROM. 4.0 did install and I just now downloaded the 4.0-to-4.1 upgrade, so I'm hoping that his will work.
What I did manage to do is come up with a pseudo-quad boot, so that I can run either ESXi, Ubuntu 10.4, Windows XP, or whatever I have on the fourth drive (it varies over short periods of time). This is accomplished via BIOS, which allows me to enable/disable SATA channels and also allows me to change the order of the hard drives that the box attempts to boot from.
I didn't even have to resort to adding drivers to the install disk (though I still have those disks). I'm hoping 4.1 installs without deleting the stuff that works now. We'll see. More later.
BTW, I've also built a 32-bit version. The only difference in the build is that you'll use "make" instead of "make amd64".
Thursday, July 15, 2010
Steps for install:
1) Go to the FiSH home page. Click on "FiSH" at the top.
2) Download the source code into a working directory. Warning! Unzipping the file will not create a sub-folder so be sure to use a working folder!
3) Download "blow.ini" and save it in ~/.xchat
4) As root, cd into your working directory and unzip the zip file that you downloaded.
5) Create a folder called MIRACL and cd into it via:
6) While still in MIRACL, download and install the Multiprecision Integer and Rational Arithmetic C/C++ Library, via:
unzip -j -aa -L miracl.zip
Note: MIRACL's author has declared the library as shareware, stating that it's free for non-commercial use. Commercial use requires a license (see the README for more info). This is why you won't see FiSH in distros like Fedora or CentOS.
7) Run the included bash script to build miracl.a. In my case, it was:
8 ) Copy the resulting miracl.a into the next higher folder and move into same, via:
cp miracl.a ..
9) Compile the plugin by running:
10) Copy the resulting plugin (xfish.so) into XChat's plugin folder via:
cp xfish.so /usr/lib64/xchat/plugins/
That's it for the code install. Using FiSH is simple, if you keep in mind that you shouldn't share keys anywhere except private chat.
Steps for Usage
1) Fire up XChat and go find a friend who's also installed FiSH. (in this example: Sparks)
2) Do a key exchange with your friend (can be open chat)
This causes a Diffie-Helman key exchange so that you can privately chat
3) Open a private chat with your friend by using the /msg+ command
/msg+ Sparks ready to generate a key?
4) Inside private chat, you'll notice that user names are inside brackets, similar to the open chat example below. You or your friend should generate a key. Ways of doing this include (but aren't limited to): manually, the Gnome Password Generator, or Steve Gibson's Ultra High Security Password Generator.
5) Share this password via secure private chat.
6) All participants should run setkey to set the key for group chat. It should look something like:
7) Enable "encrypt" and "decrypt" (they are two separate commands).
After that, your discussion should be encrypted (it should like like gibberish to anyone not having the key). Following is a screenshot showing Sparks and I having an encrypted conversation (note the brackets around the usernames).
As always, if I've made hideous errors, leave me a comment and I'll make corrections. I'll convert this to wiki notes shortly.
Sunday, July 11, 2010
While most documentation will say "live with it", it's an easy to fix problem. Switch to the Bitlbee command channel (&bitlbee) and type "rename twitter_packetgeek twitter" (substituting your info for "twitter_packetgeek", of course).
A screenshot of mine follows. Click on it to see the larger version.
- DSA generates keys faster, but verifies them slower, than RSA
- DSA supposedly scales better
- DSA doesn't have the license restrictions that RSA has
- SSH protocol 1 only supports RSA
- SSH protocol 2 supports both RSA and DSA, though many implementations only include DSA support for protocol 2
Hope this answers your question, Sparks.
Saturday, July 10, 2010
In any case, I'll start posting again once I have the old blog shut down and have it pointing here. I also need to do the minor tweaks here and wade back through 2900+ posts to get them properly categorized. The little things may move around a bit as I get settled in here.
Sunday, July 4, 2010
The extremely hard part was getting Xen to run. Ubuntu specifically supports KVM while avoiding Xen support. While a few, very talented individuals have gotten Xen to run with an Ubuntu DOM0, I'm not one of them. I'm a Xen n00b. So I went looking.
While at the South East Linux Fest, someone passed me disks for Suse Linux Enterprise 11 and OpenSuse 11.2. SLE 11 does exactly what I currently want/need. The probelm is that I can't justify buying the Enterprise license for home use.
Fortunately, OpenSuse 11.2 supports most of my requirements with a marginal amount of trouble. If you're following the wiki, you'll notice updates to various articles as I update them to reflect changes required for OpenSuse. Example: Setting up LIRC for MythTV with a MCE USB remote on OpenSuse 11.2.
For those with access to my system, I'll be working to get the background services back online. For now, OpenFire, SparkWeb, and PmWiki are back up. The rest will show up as I get the time to work on them.
Saturday, June 12, 2010
Sat in on the unofficial Fedora Activity Day, where Ian, Eric, and Paul hashed out some of the CMS and editing issues.
Thanks to Nathanial from ZenOS for buying us lunch. Looking forward to your talk.
Right now hanging out in one of the downstairs hallways, taking advantage of what little free wi-fi there is. Looking forward to the talks tomorrow.
Saturday, June 5, 2010
Saturday, May 29, 2010
Sparks and I are headed out early to take advantage of the Friday and Sunday events. We should be rolling in late Thursday night (it's a 7-hour drive). This year, maybe I can stick around and convince Ian to redo his "buiding a package" tutorial. (That was Ian, wasn't it?)
Monday, May 3, 2010
- the ability to use a second video card
I can only guess that nVidia needs to redo their drivers to support 10.4. This loss is painful in that the second card is the one with TV out, for MythTV. I'm looking at two work-arounds. Switch to using the second card as the primary display or use the MVP box again. I'll probably try the former first as the MVP option is quite limited.
The loss of Alfresco could have been considered catastropic if I hadn't backed up my data first. I was using the Alfresco as the data store for PmWiki. This allowed me to attach objects (PDFs, PPTs, DOCs, etc.) to the wiki and also have full-text search for each. The trick for doing so was using davfs2 to mount Alfresco's WEBDAV service as a file system under PmWiki. It worked nicely, though it was a bit of a kluge and required manual mounting.
Upon completion, Alfreso would start and then kill itself a few minutes later. As I'm horrible at troubleshooting Java, I only tried to fix it for a couple hours.
After that, a short search revealed Namazu, which I'm currently putting through its paces. For me, it has two plusses: it's written in Perl (something I understand) and it doesn't affect the source object that you point it at.
If I can find some time (not this year!), I'll probably play with the underlying code. Wish me luck!
Wednesday, April 21, 2010
- I follow too many people on Twitter
- one side effect of the Jabber interface is that you can tell when your friends are checking in on Facebook
So far, I've used Bitlbee and JIRC to connect various services together: Facebook, Twitter, OpenFire, Sparkweb, and a handful of IRC servers and channels. I'm currently looking at how to add OTR to jabber.
In any case, I'll push my notes to the external wiki shortly.
Monday, April 19, 2010
Sunday, April 18, 2010
Saturday, April 17, 2010
Yeah, I don't talk much but I'm still amazed by any device that lasts that long on a single charge. Only today did it start acting up because of low charge.
About the only thing that I want to change on the radio is the antenna. The stock rubber duck antenna requires that I jack the transmit power to high (~5W) so that I can reach the local repeater.
Sunday, April 11, 2010
Thursday, April 1, 2010
I've updated the howto. You can now employ Yahoo Pipes to pipe just about any feed through Google Translate to your favorite RSS reader. I'm no expert at it and there's probably a better method for doing it, but it works. There's enough detail in the howto to build your own.
Thursday, March 25, 2010
Friday, March 19, 2010
Problem was, before releasing it, he'd used a secure wipe utility on it. When I attempted to add the drive as a second datastore, ESXi balked because there was nothing on the drive, not even the partition table.
Fixing this involves running a drive partitioning tool of some sort. Luckily, I had a copy of the Ubuntu Live CD ISO in my datastore. By creating a temporary VM and booting from the CD, I was able to partition and format the drive, all without having to reboot the host machine.
In short, if you can afford the space, it's a good idea to keep your various source ISOs and tools ISOs in a folder in your datastore. It's probably not a bad idea to have the burned CDs handy too.
Thursday, March 18, 2010
Sparks! This is your fault!
Tuesday, March 16, 2010
For generic host configurations, CentOS works nicely. My problem is that I only have the one system and I'm unwilling to give up specific features/services.
In any case, I've moved back to 10.4 and am currently playing with KVM. I have a number of dedicated guest machines running and am currently learning how to grow qcow2 images.
This second part has been a bit difficult in that qemu has some quirks when it runs live CDs (such as qparted which is needed to resize a partition inside of an imaged filesystem). An example of this is where, while running the qparted live CD under qemu, the mouse "sticks" to the right-hand side. It took a little searching but this issue was overcome by running "export SDL_VIDEO_X11_DGAMOUSE=0" on the command line just before starting qemu. (Hint: once you're done with qparted, you'll want to close the terminal window to destroy that export.)
In any case, KVM seems to be meeting most of my needs. I'll start adding notes to the wiki shortly, in with what's already there for ESXi, Virtualbox, and Xen (ouch).
Sunday, February 21, 2010
Monday, February 15, 2010
- my hardware isn't young enough to support native virtualization
- OS virtualization doesn't support mixed environments
- full virtualization tends to bog down on my older equipment (VirtualBox, VMware)
- ESXi is overkill for my hard ware (need a second box for it to be useable)
- KVM is a bit too light
That leaves Xen. I picked up "The Book of Xen" at the recent ShmooCon and have played with the CentOS version enough that I want to try the Debian version and play with DTC-Xen. If it allows access to some of my not-normally-supported-in-VM hardware (and it looks like it will), I'll probably make the switch permanent.
Saturday, February 13, 2010
Trying to figure out how to (safely) mention S.E.L.F. to wife. Gained some spouse points by putting together an intracate "some assembly required" desk for her. Torn between mentioning next conference or need for 2-meter transceiver.
Hope she's not reading this.
Sunday, January 31, 2010
For those that don't recognize the tool, vmfs-fuse allows you to mount VMFS formatted volumes under "regular" linux. I was backing out of an extremely large (1TB) ESXi install and, half-way through, I realized that I hadn't backed up one of the virtual machines. I ran up against a wall, trying to get the usual disk utilities to mount the volume and was looking at having to juggle hard drives and reinstall ESXi, just to get a folder copied. A search of Google and a Synaptic search later, and I'm copying the folder as I write this post.
So, if you're in a pinch and need to avoid having to install ESXi, just to get a file/folder copied, you might want to consider installing the vmfs-tools package (which contains vmfs-fuse).
Tuesday, January 26, 2010
Monday, January 25, 2010
Sunday, January 17, 2010
Short version: after about 6 hours of trial and error, we have a working install.
Longer version: Took advantage of work already done by Varazir and geppi (in the VM-Help forum). In the course of six hours, we saw both the lvmdriver error and the pink screen of death. Couldn't get the USB install to work but did get an ISO built that successfully installed ESXi on the box. Noticed Hauppauge hardware info in the psid listing that makes me want to retry attempting to access the PVR-250 card via ESXi.
In any case, notes are in the wiki. I'll work on adding graphics. Enjoy!