i originally wrote the following on 23-may-2007:

i installed red hat enterprise linux 4.5 on an hp dl365 yesterday to test out iscsi.

we carved out some space on the san to use for the testing, 500gb to be exact. the dl365 has a qlogic fiber-channel hba in it, connected to the hp san at 4 gbps. the 500gb of storage shows up on the rhel box at /dev/sda. no partitions or filesystems were created on the device.

as far as i can tell, rhel does not include support for being an iscsi target, which i did not find out until after i had it installed. fortunately, i came across the iscsi enterprise target project on sourceforge. their wiki led me to martin’s “iscsi target howto on enterprise linux (rhel4)“. by following that, i was able to get the iscsi target up and running, exporting the 500gb on /dev/sda.

the next step was to connect to that storage space from a windows box. this test is sort of a proof-of-concept to see if we can get things to work the way we want — which means windows “clients”, or initiators, will be used. i found anze vidmar’s “going enterprise — setup your fc4 iscsi target in 5 minutes” wiki page, which details setting up an iscsi initiator on windows. i grabbed the microsoft iscsi software initiator version 2.04 and installed it on my windows xp workstation (a vista version wasn’t available, or i’d have went for that).

following anze’s instructions allowed me to get the windows xp client configured as an initiator in just a few moments, and i had a p: drive showing up as a local disk, with an ~500gb ntfs filesystem on it for all my storage pleasure. excellent!

unfortunately, we need some access controls in our environment. if or when this goes into production, all iscsi traffic will be on an isolated, private network. i’m a big fan of the layered security approach, however, so while an isolated, private network is a good start, i want to implement the authentication that iet and the microsoft initiator are supposed to support.

enter problem. =)

anytime i try to set up some credentials on the target side (using “incominguser username password” in the /etc/ietd.conf config file) and use those same credentials in the microsoft iscsi initiator, i get a simple “authentication failure” dialog box on the client/initiator side. unfortunately, there aren’t any log entries on the server/target side (that i noticed, anyway) to help provide any insight.

anyone ran into this before and have any suggestions? tia.

update: seems i didn’t have any credentials listed in the “global” section of the /etc/ietd.conf file, which is needed if you try to do authentication during the discovery phase (i was). added that in and now have authentication working across the board.

a month or so ago, i downloaded jungle disk. i installed it on an older laptop i have running windows xp that has quite a few files on it i’d like to have backups of. i uploaded a few hundred megabytes of files to disk storage on the amazon s3 system and pretty much forgot about it.

a few days ago i was looking at my bank statement and seen a charge from amazon for $0.04 for the s3 storage. that got me thinking about backups again. i don’t really have any good backups here at home, other than keeping copies of “important stuff” on multiple machines — not exactly an ideal backup strategy.

i have an external usb drive that i’ve been wanting to use with time machine on my macbook to keep it backed up (it has pretty much became my “primary computer” in the last few months). i was keeping copies of important files on it, though, so i couldn’t just wipe it clean. instead, i first hooked it up to an ubuntu linux box that i have here, copied off all the important stuff, and then wiped it clean.

the external usb drive is now hooked up to the macbook and the first full backup (using time machine) is taking place as i write this (in textmate, of course!). it’s currently at 7.7gb of 105.9gb to back up. fun!

on the ubuntu linux box, jungle disk is hard at work backing up all my important files to the amazon s3 storage. i started out with 2.4gb of data to upload, mostly pictures and video that wouldn’t be easily replaced. with my cable modem connection (10 mbps down/1 mbps up), it’s going to take a while. once i get all the “important stuff” dumped onto amazon’s servers, however, and only occasionally add stuff to it, the backups will run much quicker.

i still have a few more pc’s to sift through in order to find everything i want backed up, but i don’t anticipate having more than 10gb or so of data stored on amazon’s servers. at their rates, that makes for very cheap off-site backups.

if you’re looking for a good solution to this same problem, i highly recommend amazon s3 and jungle disk. i’ve had no problems with either as of yet (granted i haven’t used either extensively). jungle disk is commercial software (only $20), but you can run it on as many pc’s as you have. they also have windows, linux, and os x versions that operate nearly identical. i haven’t installed the os x version yet, but i imagine i will before the day is over.

windows toaster

the windows toaster looks great, but sometimes it just won’t make toast. it either comes out burnt or raw, and you have to unplug the toaster and plug it back in again each time you want to try and make some toast. for every loaf of bread you buy you are forced to buy a new toaster to go with it.

linux toaster

the linux toaster looks absolutely awful: it has wires crimped together, things are just hanging out of it. the first time you make toast with it the toaster burns it; the next time it’s raw. you read the man pages and invoke the command line “toast -verbose -breadsize 50132 -eject -o z3321 > /dev/toast” and it makes perfect toast ever after.

mac toaster

the mac toaster has no settings or controls. it looks very stylish, but will only accept proprietary-sized bread which can only be bought from apple dealers at ten times the cost of regular bread. the toast is fine except that the size of the bread is so odd that you can’t actually eat the toast it produces, although it does look very good.

just wanted to brag real quick that one of my debian gnu/linux servers surpassed the one year of uptime mark over the weekend:

14:25:57 up 366 days, 21:29, 0 users, load average: 0.00, 0.00, 0.00

let’s see a windows server box do that. =)

chris howard wrote an opinion article on apple matters called “apple is killing linux on the desktop”. in the article, chris talks about how the number of apple macs running os x used to access the web has been consistently increasing over the last few years while linux has seen only minimal growth.

as a long-time linux user and recent mac convert, i can’t help but feel that i’m helping this along. i began using linux sometime in 1997. since then, i’ve never been able to use linux exclusively. in other words, i’ve always had to keep around at least one windows box in order to do certain tasks. that may just be changing, however.

less than two weeks ago, i got my first mac. i say “first mac” because, even this early in the game, i’m fairly confident it won’t be my last. i’ve even went so far as to tell myself that i’ll never buy another windows pc again. that may seem a bit premature, but i’m pretty convinced.

i work in information technology. there aren’t many days that go by that i’m not either sitting in front of a computer or have one on my lap for at least 12 hours. i am, by all accounts, a “power user”. by day, i manage computer networks. my group runs the servers that keeps our organization going. if i didn’t require outlook, i could use a linux pc 90% of the time and be just fine. in fact, i have both windows xp and linux pc’s under my desk. until two weeks ago, however, i worked almost exclusively on the xp box.

i write a fair amount of code. it’s not my official job but if there’s applications that we can create or scripts we can write to make our jobs easier, that task usually falls to me. i primarily use perl for unattended scripts and php for our web-based apps (cakephp rocks, by the way). i also tend to carry a laptop back and forth to work with me. after a certain point, it became a huge pain in the ass to use my laptop to write code but have that code actually reside on another server. if i’m at work, it meant ssh’ing to the box my code was on (i use vi for 90% of my code development). if i’m at home (or somewhere else), it meant bringing up the vpn link and then doing the same thing.

i was able to avoid that by running linux on my laptop and just developing all my code there. then, however, i was missing out on tools that i use quite often (e.g. ms word, ms outlook, etc.) so if i needed to use them, i still had to connect to a windows box to use them.

i managed to convince my $boss to buy me a mac. one of my selling points was that it would make writing code easier — and, as an extension, faster — because all the development tools are right there. as i write this on my new macbook, i can quickly open up a terminal and i have instant access to both perl and php. in addition, i’ve got apache and mysql running. that’s everything i need to write and test my code locally. i can do it anywhere, with or without a network connection. that’s nice too, since we travel a lot and it means i can write code while riding down the highway (i’ve already done it).

anyways, i’m getting a little sidetracked. my point is that i can agree with chris. previously, if you were looking for an alternative to windows, you went with linux. now more and more people are switching to apple and os x instead. i’ve been telling everyone that i get the best of both worlds (windows and linux) on this mac. i haven’t been able to find one single thing that i can do on either windows or linux that i can’t do on this mac. this macbook does everything, and does it quite well.

i’ve still got my windows xp and linux boxes, both at work and at home. that may very well change in the future.