So, I was just thinking, what if I had two 80 GB HD's and a 160 GB HD? I could then gstripe the 80 GB's, and then on top of that stripe those with the 160 GB. Then I'd have around 320 GB or so, I could then gstripe that with a 320 GB HD.
Ohhh the possibilities. By that time, I'd probably just use some sort of gvinum, since apparently it can do raid 5 over hard drives that are not the same size. Something i'd like to eventually do as a way to keep my data safe and a secure.
I was really bored. Which normally results in some destructive thing happening to something I am playing with. This time it was Some of the hard drives in my FreeBSD box. Just a little back-story:
/dev/ad0 - 20 GB /dev/ad1 - 40 GB /dev/ad2 - 100 GB /dev/ad3 - 60 GB /dev/da0 - 8 GB /dev/da1 - 8 GB
So, I had already made a gstripe(8) out of the two SCSI drives, da[1|2], however I wanted to see if I could do the same with ad[1|2|3]. As I did not want to resort to gvinum, which is rather heavy, and I was unable to figure out how it worked. The entire point behind gstripe is that since it is striped, the writes/reads are faster since it writes/reads to all the drives at the same time. So first I tried:
gstripe label -v st1 /dev/ad1 /dev/ad2 /dev/ad3
That however complained that it would only use 40 GB's on all of the drives, since the smallest drive in the collection was 40 GB. I then read up on gconcat(8), which concatenates all the drives together to look like one big drive, without the speed improvement of striping. So I figured, since FreeBSD's gstripe is GEOM based, it should not care about what the underlying data sink is, just that it is a device, so just maybe this could work:
gconcat label -v gb100 /dev/ad1 /dev/ad3
So now I had created a 100 GB "hd" at /dev/concat/gb100. I now did what I was planning to do:
gstripe label -v st1 /dev/ad2 /dev/concat/gb100
No problem! :-D. I now had 180.48 GB, of which only about 166 GB was usable. Which is still excellent. As it still came with a speed improvement. Followed up with the standard advice from the handbook I did the following:
bsdlabel -wB /dev/stripe/st1 newfs -U /dev/stripe/st1a mount /dev/stripe/st1a /mnt
Which got me the following:
/dev/stripe/st1a 180G 5.5G 161G 3% /mnt
Full:
Filesystem Size Used Avail Capacity Mounted on /dev/ad0s1a 496M 63M 393M 14% / devfs 1.0K 1.0K 0B 100% /dev /dev/ad0s1e 496M 230K 456M 0% /tmp /dev/ad0s1f 15G 2.2G 11G 17% /usr /dev/ad0s1d 1.4G 115M 1.2G 9% /var /dev/stripe/st0a 16G 4.0K 15G 0% /usr/scsi /dev/stripe/st1a 180G 5.5G 161G 3% /mnt
w00tage. Next up I might actually add gdbe(8) or geli(8) even in front of the stripe, so that I will have an encrypted stripe. Talk about stacking. Might give gvinum a try sometime :P
All working:
keyhole# gconcat status Name Status Components concat/gb100 UP ad1 ad3 keyhole# gstripe status Name Status Components stripe/st0 UP da0 da1 stripe/st1 UP ad2 concat/gb100
I am moving my domains away from registerfly.com (not linked on purpose) and am moving them to GoDaddy. I was sick and tired of RegisterFly's website loading slower than molasses coming down from a mountain top. The inability to keep my account secure (someone logged into it, I got an email, and immediately change the password).
I apologize if there are any domain hickups in the process. I hope this helps me in the future for when I need to work with my domains, and change the DNS settings. Waiting for the login to happen for over 40 seconds is just not acceptable.
Wanna play with a christmas tree remotely? Turn it on and off in our lobby? Here is your chance, visit http://uatree.0x58.com/index.html to turn the tree on, off, or scroll the lights up and down. And maybe, if the camera is pointing right, you can catch a glimpse of it on there!