OpenBSD CVS RAID Array Failing, Needs Replacement

Sam writes "The OpenBSD cvs server has a failing RAID array. Users of the projects on that array: OpenBSD, OpenSSH, OpenBGPD, OpenNTPD, and the upcoming OpenCVS are all invited to contribute towards the $12,500 cost of a suitably high-spec replacement. OpenBSD Journal article, and original request (thread)."

Gee..

[brilinux (255400)]

I guess that it is a good thing that I decided to spend all day today compiling NetBSD instead of OpenBSD ... but, those projects are somewhat important, especially OpenSSH; if I were not a poor college student, I would contribute. Good luck.

Re:Gee..

[Noksagt (69097)]

if I were not a poor college student, I would contribute.

Philanthropy does not require you to be personally wealthy. For one, a small donation from you of $5 will almost certainly help them & not overly burden you. But, as a student, you are in a good position to actually fundraise for them as well. Not only can you encourage your peers to also donate $5, but you can solicit faculty, alumni, and departments to donate much more than $5.

OpenBSD also accepts hardware donations. You can send any spare equipment you have, encourage others to do the same, and/or even dumpster dive for perfectly working components that could use a new home.

OpenBSD's RAID is Dying

[NitsujTPU (19263)]

It is official; Netcraft confirms: OpenBSD's RAID is dying

One more crippling bombshell hit the already beleaguered OpenBSD community when IDC confirmed that the OpenBSD RAID has failed again, now down to less than a fraction of 1 percent properly operating. Coming on the heels of a recent Netcraft survey which plainly states that OpenBSD's raid has lost more sectors, this news serves to reinforce what we've known all along. OpenBSD's RAID is collapsing into complete Redundant Disarray of Inexpensive Disks, as fittingly exemplified by failing dead last [samag.com] in the recent Sys Admin comprehensive networking test.

You don't need to be a Kreskin [amdest.com] to predict OpenBSD's RAID's future. The hand writing is on the wall: OpenBSD's RAID faces a bleak future. In fact there won't be any future at all for OpenBSD's RAID because OpenBSD's RAID is dying. Things are looking very bad for OpenBSD's RAID. As many of us are already aware, OpenBSD's RAID continues to data. Red ink flows like a river of blood.

Fact: OpenBSD's RAID is dying

Link to make a donation

[abradsn (542213)]

How do we actually make a donation?

Donation Method from TFA

[Noksagt (69097)]

"If you want to help, paypal some money to slash at peereboom dot us, or you can use the OpenBSD ordering system. Be sure to mention its for the cvs machine."

Expensive

[BoomerSooner (308737)]

It would seem to me they could build one for significantly less. Dell's 4 hour service isn't as fast as having spare parts on-hand and swapping them yourself (someone has to be there to let Dell in so why can't they pull a harddrive sled and slide in a new one?). Plus the savings by building it themselves would more than cover the price of spare harddrives/controllers, etc...

That is just my experience. Dell's service/support is pretty good but I've had a significantly higher rate of failure on their hardware compared to purchasing components individually.

Re:Expensive

[SunFan (845761)]

Yes, but would you want something as important as these projects running on a home-built unit? They need something that's actually known to work beforehand.

Re:Expensive

[Kalak (260968)]

The cost of a spare backplane for the enclosure on this level of equipment is more than a few dollars. We're not just talking having a spare power supply and drives here. 4 hour service is a deal breaker when dealing with highly critical equipment of this level. No amount of spare parts, short of complete redundancy in equipment (another $12k) would be enough to cover every contingency. It's one reason our shop is not going to get an XRAID form Apple - good price point, but our critical infrastructure needs more than next business day response and some "common" spare parts. We're planning on even buying the redundant RAID array as well. Good Sysadmins are supposed to be paranoid and ready to face major disaster with excitement, and to plan for that disaster that should never occurr.

Re:Expensive

[OrangeSpyderMan (589635)]

I agree. At one point Dell sold the company I work for some servers that would be clustered and hooked up to SAN's the world over. They sold us "gold" maintenance, knowing fine well where each and every cluster would be located, because we told them, and they even factored that in to the overall quote. This isn't some little deal for a couple of tens of thousands of dollars, this is a 80,000+ users worldwide, who would all be moving from HP desktops and servers to Dell, for the Wintel stuff. Now guess what nearly blew the deal? One of the locations that was destined to receive a Dell cluster was not supported for 2-3 days a week because their support engineer organised deep-dea fishing trips!!! He was the Dell engineer, and they came within a hair's breadth of blowing this as they furiously scrambled for engineer cover in $LOCATION. I wouldn't give Dell ANY money for service over their default baseline, as they may know how to ship boxes, they don't know jack shit about service.

question for you guys.

[by Anonymous Coward]

What does the 'i' stand for in RAID?

Thanks.

Independent

[Noksagt (69097)]

While this is probably a dig at the cost of the project, many now say the 'i' stands for 'independent' rather than the original 'inexpensive.' There are good reasons to buy very large, high quality SCSI disks for a project like this.

Current Configuration & Need to Upgrade

[Noksagt (69097)]

From this post [theaimsgroup.com]:

The current raid array is using 14 U160 drives in a dual raid5

configuration with a couple of hot spares. It is time to build up.
See, the actual cvs machine is just a p3/gig machine. While there are
lots of much faster build machines on the network, there has been no
reason to crank the processor on cvs. It is not cpu bound but
*strictly* IO bound. And the raid configuration has been working very
well to keep that IO load under control, well it has been kind of
working.

We support more architectures, which means more NFS load is being
generated. There are more developers, and that does affect things
because developers working fast do checkouts directly off cvs instead
of via mirrors. As well, the array is full (it is half 18GB drives
and half 36GB drives now due to failures after replacement). All of
them are U160 drives.

Now there appears that the raid backplane is developing some issues,
and at the same time, it is time for the "every three to four years"
replace some parts plan. It's what most IT shops do as well. The
drives are also getting a bit up there in age now. Perhaps that is
why I am starting to lose them more often.

But if we want more oomph, then it is time to go to U320. It is also
time to move towards another raid controller (already have it) which
we hope will have supported raid management soon.

shut the damned thing down now...

[advocate_one (662832)]

while you've still got good data... take a backup first though... but shut it down...

Well, okay, then...

[jawtheshark (198669)]

I already donate yearly... Every november I buy the latest CD and substract the price from 100Euro. That part is simply donated. It's not much, I know, but I'm just an individual. Yes, and ever for such small amounts you make it to the donation page. Cool, eh? (This also means your name is printed on the sleeve of the CD)

I'll make an exceptional donation... I use OpenBSD on so many systems (now even on a SMP systen... yay!) that I owe Theo and Co.

Not just users of those projects

[Bastian (66383)]

If you use any OSS unix-like, or many OSS tools other than something with an Open* name, you are likely using at least a few things that have benefitted directly from the OpenBSD project. In an effort to keep OpenBSD secure, they contribute security patches to all sorts of software that runs on OpenBSD.

In particular, I'd encourage everyone who uses Linux to contribute.

Re:Not just users of those projects

[by Anonymous Coward]

Well, OpenSSL, OpenGPS, OpenGFS, OpenOffice, OpenVB, OpenTTD, OpenVPN and OpenWFE just to name a few aren't actually associated with OpenBSD or even BSD licensed. In truth, most Open* things are under the GPL and done by other people. The good OpenBSD stuff is pf, OpenSSH, OpenBGP, OpenNTP, OpenCVS and OpenBSD itself.

Apple???

[fdawg (22521)]

I thought Apple based OSX on BSD. Why aren't they flying to BSD's aid? Im actually quite surprised more big businesses aren't riddling the BSD community with money. I realize the whole OSS movement is built on the philosophy of "share and share alike", but blindly ripping the work of others without giving back is gross.

It is done.

[Nimrangul (599578)]

The money has been raised, the purchase shall soon be made. The link is here [theaimsgroup.com] and you will note that the only companies that put in any money are smaller ones and the rest of the money has come from individuals.

Re:$12,500... what is this, MAC hardware?

[by Anonymous Coward]

I think it's because they want quality equipment with a service contract. Your RAID enclosure built of duct tape and popsicle sticks fails to amuse.

Re:$12,500... what is this, MAC hardware?

[Triumph The Insult C (586706)]

hmm, let's look at the facts

the person organizing the replacement (marco@) works with at lowest level of disks as a day job. he is very involved in all of the obsd scsi stuff

you agree that scsi is "outdated super expensive technology", but offer no facts, nor have you bothered to check the configuration they currently have

who should i consider more knowledgable in this subject matter?

for the kind of work that cvs.openbsd.org does, ide simply will not cut it

for storing big pr0n movies, ide works fine. for lots of cvs commits and checkouts and heavy i/o from nfs, ide sucks

Re:$12,500... what is this, MAC hardware?

[Noksagt (69097)]

From TFA:

The price of a fully populated PowerVault 220S with 4 hour on-site warranty is

about $12500 USD including tax and shipping. Let me kick this off by donating
$250 to this cause. A person that shall not be named donated a PERC4/DC RAID
controller.

Re:Okay,

[Homology (639438)]

if I had $12,500 I'd have to offer it as a bounty for an original *BSD is dieing troll. The variation on a theme are just not as funny as they once were.

It's well know that the troll community is suffering greatly from excessive inbreeding. It's really noticable that this inbreeding is seriously impairing their already scant intellectual skills. They are not even capable to do a simple copy/paste/replace/post. They'll quite simply breed themselves to extinction, so you can donate that bounty to OpenBSD instead.

Re:"Open" CVS?

[drinkypoo (153816)]

Try "less open" CVS. CVS is GPLed.

Arguably, GPL software is more Free (with a capital F) while BSD software is more Open. I say arguably because people have been arguing about this shit on slashdot for ages (myself included.)

Re:"Open" CVS?

[Nimrangul (599578)]

FSF views anything they don't make as less free, regardless of anything you may say or think.

Also, the FSF believes everyone has the right to all code, thus anything that allows for a closed source version of anything denies the "right" to said code and is therefore bad.

Another note: The OSI are irrelevant, the only thing that matters for making something open source is there being access to the source; a little slip of words mean nothing, nor does the "thumbs up" from an organisation that does not contr

Re:Perhaps...

[Nimrangul (599578)]

You may not have heard about this, so I'll let you in on a secret, Compaq is dead. HP had a merger with them and now controls all that was Compaq.

This would make it slightly difficult to buy from them.