Re: big machines running Debian?
Alex Samad a...@samad.com.au writes: On Sat, Feb 28, 2009 at 09:50:06AM +0100, Goswin von Brederlow wrote: Alex Samad a...@samad.com.au writes: [snip] true, depends on whos rule of thumb you use. I have seen places where mandate fc drives only in the data center - get very expensive when you want lots of disk space. The only argument I see for FC is a switched sorage network. As soon as you dedicate a storage box to one (or two) servers there is really no point in FC. Just use a SAS box with SATA disks inside. It is a) faster, b) simpler, c) works better and d) costs a fraction. The problem I have seen is the person who controls the purse strings doesn't always have have the best technological mind. There was a while back where have fibre meant fibre to the disk. So managers wanted fibre to the disk, so they paid for fibre to the disk. And now they have to learn that we have new technologies. New requirements and new solutions. What was good 5 years ago isn't neccessarily good today. Saddly enough a lot of purse strings seem to be made of stone and only move in geological timespans. :) And hey, we are talking big disks space for a single system here. Not sharing one 16TB raid box with 100 hosts. Also the disk space might not be need for feeding across the network, db aren't the only thing that chew through disk space. the op did specific enterprise, I was think very large enterprise, the sort of people who mandate scsi or sas only drives in their data centre They have way to much money and not enough brain. I would have to dissagree, some times the guidelines that you set for your data storage network mandate having the reliability (or the performance) of scsi (or now sas), they could be valid business requirements. Could be. If you build storage for a DB you want SAS disks and raid1. If you build a petabyte storage cluster for files 1GB then you rather want 3 times as many SATA disks. An XYZ only rule will always be bad for some use cases. Traditionally scsi drives had a longer warranty period, were meant to be of better build that cheap ata (sata) drives. Although this line is getting blurred a bit. There surely is a difference between a 24/7, 5 year warranty, server SCSI disk and a cheap home use SATA disk. But then again there are also 24/7, 5 year warranty, server SATA disks. I don't think there is any quality difference anymore between the scsi and sata server disks. Unless we talk about a specific situation, storage as other areas of IT are very fluid, and there are many solutions to each problem. Exactly. Look at the big data centers of google and such that use pizza box's machine dies who cares its clustered and they will get around to fixing it at some point. to 4-8 nods clusters of oracle that are just about maxed out, one server goes down and Same here. Nobody builds HA into a HPC cluster. If a node fails the cluster runs with less node. Big deal. Saddly enough for storage there is a distinct lack of software/filesystems that can work with such a lax reliability. With the growing space requirements and stalling size increase in disk size there are more and more components in a storage cluster. I feel that redundancy has to move to a higher level. Away from the disk level where you have raid and towards true distributed redundancy across the storage cluster as a whole. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 28, 2009 at 10:14:15AM +0100, Goswin von Brederlow wrote: Hot-spare devices work just fine (see below). What doesn't exists afaik are global hot spares. E.g. 7 disks, two 3 disk raid5 and one spare disk for whatever raid fails first. You would have to script that yourself. Any idea what the spare groups mentioned in the mdadm documentation are? I would have guessed it was something to do the global spare thing. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 28, 2009 at 10:16:19AM +0100, Goswin von Brederlow wrote: Not to repeat myself but a GPT with an entry for /boot in its fake MS-Dos table works just fine. Perhaps, but why bother. Just using GPT works. And it won't confuse any tools that actually know how GPT is supposed to be used. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 28, 2009 at 09:56:04AM +0100, Goswin von Brederlow wrote: Ron Johnson ron.l.john...@cox.net writes: On 02/25/2009 03:48 PM, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Ah. The minicomputer tradition I come from (and thus how I organized my home PC) is to have a relatively small OS/swap disk and a separate data array. Of course, max device size always gets bigger, and smaller devices fall off the market... I'm aiming for a small SSD disk for the system and seperate data array that can be spun down most of the time. It doesn't take much with modern SATA drives to hit 2TB. Given we can get 1.5TB in a single drive, how many months before we can get 2TB in a single disk. Later this year. Waiting for them. 1.5TB disks don't mix so well with 1TB disks in raid. I don't want to split the disks into 0.5TB partitions and then raid over those. Anyone know how/if windows copes with 2TB disks? Does it understand GPT too? By the looks of it, it supports GPT for data drives, but is incapable of booting from it (except itanium versions where it is the default format) And only 64bit versions of XP and 2003 server appear to support it, while vista and 2008 server seem to always support it (but not for boot unless your machine has EFI). It seems microsoft belives you need EFI to boot from GPT, even though grub2/linux has no such limitation. Go figure. Of course a 2TB disk is OK with MBR still. Anything larger will have a problem though. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 28, 2009 at 09:39:07AM +, Ian McDonald wrote: Erm, not on anything other than a sequential read (and even then, I've never seen a single disk that would actually sustain that across it's whole capacity). Even raid-5s of significant numbers of disks aren't enormously fast, especially under multiple access. hdparm informs me that the SATA 28+2 spare raid-5 I have will read 170M a second. That would rapidly diminish under any sort of load. Probably a bus limitation rather than the disks. I have no problem getting 100MB/s on a 4 disk raid5 with SATA. The only thing we've found that'll stand up to real multiuser load (like a mail spool) is raid-10, and enough spindles. We're beginning to see the requirement for 10GE on busy machines. Sure. All depends on the load after all. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Lennart Sorensen wrote: On Sat, Feb 28, 2009 at 10:14:15AM +0100, Goswin von Brederlow wrote: Hot-spare devices work just fine (see below). What doesn't exists afaik are global hot spares. E.g. 7 disks, two 3 disk raid5 and one spare disk for whatever raid fails first. You would have to script that yourself. Any idea what the spare groups mentioned in the mdadm documentation are? I would have guessed it was something to do the global spare thing. I haven't tested shared hot spares myself but this article (among quite a few other hits from google: shared hot spare) indicates that it is quite possible with mdadm running in daemon mode: http://www.redhat.com/magazine/021jul06/departments/tips_tricks/ ...and it appears Len was completely right about the spare-group keyword. Cheers, Jonas -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Mon, Mar 02, 2009 at 02:28:04PM +0100, Goswin von Brederlow wrote: Alex Samad a...@samad.com.au writes: On Sat, Feb 28, 2009 at 09:50:06AM +0100, Goswin von Brederlow wrote: Alex Samad a...@samad.com.au writes: [snip] And now they have to learn that we have new technologies. New requirements and new solutions. What was good 5 years ago isn't neccessarily good today. Saddly enough a lot of purse strings seem to be made of stone and only move in geological timespans. :) some do and some don't - the old ideology of nobody got sacked for buying IBM [snip] Could be. If you build storage for a DB you want SAS disks and raid1. If you build a petabyte storage cluster for files 1GB then you rather want 3 times as many SATA disks. An XYZ only rule will always be bad for some use cases. True I had a customer buy a 8T fc disk array lustre based and then they expanded it soon afterwards Traditionally scsi drives had a longer warranty period, were meant to be of better build that cheap ata (sata) drives. Although this line is getting blurred a bit. There surely is a difference between a 24/7, 5 year warranty, server SCSI disk and a cheap home use SATA disk. But then again there are also 24/7, 5 year warranty, server SATA disks. I don't think there is any quality difference anymore between the scsi and sata server disks. Unless we talk about a specific situation, storage as other areas of IT are very fluid, and there are many solutions to each problem. Exactly. Look at the big data centers of google and such that use pizza box's machine dies who cares its clustered and they will get around to fixing it at some point. to 4-8 nods clusters of oracle that are just about maxed out, one server goes down and Same here. Nobody builds HA into a HPC cluster. If a node fails the cluster runs with less node. Big deal. you would be surprised how many people want HA head nodes Saddly enough for storage there is a distinct lack of software/filesystems that can work with such a lax reliability. With the growing space requirements and stalling size increase in disk size there are more and more components in a storage cluster. I feel that redundancy has to move to a higher level. Away from the disk level where you have raid and towards true distributed redundancy across the storage cluster as a whole. yes it would be nice. My thoughts are we haven't seen any big jumps in data storage for a while, nothing like we are seeing in memory and cpu speed. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- Our enemies are innovative and resourceful, and so are we. They never stop thinking about new ways to harm our country and our people, and neither do we. - George W. Bush 08/05/2004 Washington, DC signature.asc Description: Digital signature
Re: big machines running Debian?
lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Sat, Feb 28, 2009 at 10:16:19AM +0100, Goswin von Brederlow wrote: Not to repeat myself but a GPT with an entry for /boot in its fake MS-Dos table works just fine. Perhaps, but why bother. Just using GPT works. And it won't confuse any tools that actually know how GPT is supposed to be used. Because then grub works. :) MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Mon, Mar 02, 2009 at 08:32:22PM +0100, Goswin von Brederlow wrote: lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Sat, Feb 28, 2009 at 10:16:19AM +0100, Goswin von Brederlow wrote: Not to repeat myself but a GPT with an entry for /boot in its fake MS-Dos table works just fine. Perhaps, but why bother. Just using GPT works. And it won't confuse any tools that actually know how GPT is supposed to be used. Because then grub works. :) But grub2 works when you use a GPT partition table that is actually compliant with the standard. I am pretty sure your setup is NOT GPT comliant. It just happens to work. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 03:21:44PM -0600, Ron Johnson wrote: On 02/26/2009 01:49 PM, Ron Peterson wrote: 2009-02-26_14:21:54-0500 Douglas A. Tutty dtu...@vianet.ca: On Wed, Feb 25, 2009 at 08:53:45PM -0600, Ron Johnson wrote: On 02/25/2009 07:22 PM, Douglas A. Tutty wrote: [snip] /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO Why is Read Ahead disabled on Logical Drive 1? My understanding is that read ahead in this case refers to the ability of the raid card to read ahead from one disk while a read is taking place on another disk. This only makes sense in a redundant raid level. LD1 is raid0, so there is no other disk from which to read ahead. My understanding is that read ahead means the controller reads more data into memory than you asked for, expecting that the next bits you ask for will be immediately after the ones you just got. That *is* the standard definition. Though there's nothing stopping Megaraid from being weird. I just checke the setup in the bios and it is set for adaptive read-ahead on both LDs. I don't know what's wrong with the output from /proc. Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 28, 2009 at 09:50:06AM +0100, Goswin von Brederlow wrote: Alex Samad a...@samad.com.au writes: [snip] true, depends on whos rule of thumb you use. I have seen places where mandate fc drives only in the data center - get very expensive when you want lots of disk space. The only argument I see for FC is a switched sorage network. As soon as you dedicate a storage box to one (or two) servers there is really no point in FC. Just use a SAS box with SATA disks inside. It is a) faster, b) simpler, c) works better and d) costs a fraction. The problem I have seen is the person who controls the purse strings doesn't always have have the best technological mind. There was a while back where have fibre meant fibre to the disk. So managers wanted fibre to the disk, so they paid for fibre to the disk. And hey, we are talking big disks space for a single system here. Not sharing one 16TB raid box with 100 hosts. Also the disk space might not be need for feeding across the network, db aren't the only thing that chew through disk space. the op did specific enterprise, I was think very large enterprise, the sort of people who mandate scsi or sas only drives in their data centre They have way to much money and not enough brain. I would have to dissagree, some times the guidelines that you set for your data storage network mandate having the reliability (or the performance) of scsi (or now sas), they could be valid business requirements. Traditionally scsi drives had a longer warranty period, were meant to be of better build that cheap ata (sata) drives. Although this line is getting blurred a bit. Unless we talk about a specific situation, storage as other areas of IT are very fluid, and there are many solutions to each problem. Look at the big data centers of google and such that use pizza box's machine dies who cares its clustered and they will get around to fixing it at some point. to 4-8 nods clusters of oracle that are just about maxed out, one server goes down and MfG Goswin PS: The I in RAID stands for inexpensive. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- Of course he was all in favour of Armageddon in *general* terms. -- (Terry Pratchett Neil Gaiman, Good Omens) signature.asc Description: Digital signature
Re: big machines running Debian?
Hi, Am 2009-02-21 08:00:32, schrieb Igor Támara: Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. I am owner of three Sun Blade (Sparc) and each has 32 CPUs, 128 GByte of memory and 160 SCSI drives of 300 GByte in 10 cages. I have started in end of 1999 with it and 64 drives of 76 GByte. UPdated to 147 GByte drives and 2007 to 300 GByte ones. Sun machines are killers... and they run like the heaven. Thanks, Greetings and nice Day/Evening Michelle Konzack Systemadministrator 24V Electronic Engineer Tamay Dogan Network Debian GNU/Linux Consultant -- Linux-User #280138 with the Linux Counter, http://counter.li.org/ # Debian GNU/Linux Consultant # http://www.tamay-dogan.net/ http://www.can4linux.org/ Michelle Konzack Apt. 917 ICQ #328449886 +49/177/935194750, rue de Soultz MSN LinuxMichi +33/6/61925193 67100 Strasbourg/France IRC #Debian (irc.icq.com) signature.pgp Description: Digital signature
Re: big machines running Debian?
Am 2009-02-25 16:48:30, schrieb Lennart Sorensen: It doesn't take much with modern SATA drives to hit 2TB. Given we can get 1.5TB in a single drive, how many months before we can get 2TB in a single disk. Ehm, HOW MANY what? The 2 TByte drives are already out. Some selected customers of Hitachi have them already. (one of them is here in Strasbourg) Thanks, Greetings and nice Day/Evening Michelle Konzack Systemadministrator 24V Electronic Engineer Tamay Dogan Network Debian GNU/Linux Consultant -- Linux-User #280138 with the Linux Counter, http://counter.li.org/ # Debian GNU/Linux Consultant # http://www.tamay-dogan.net/ http://www.can4linux.org/ Michelle Konzack Apt. 917 ICQ #328449886 +49/177/935194750, rue de Soultz MSN LinuxMichi +33/6/61925193 67100 Strasbourg/France IRC #Debian (irc.icq.com) signature.pgp Description: Digital signature
Re: big machines running Debian?
On 02/28/2009 03:14 AM, Goswin von Brederlow wrote: Ron Johnson ron.l.john...@cox.net writes: On 02/27/2009 07:50 AM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 05:58:43PM -0600, Ron Johnson wrote: As would auto-replacement of bad drives by hot spares. Usually the firmware of a raid card does that itself. If a drive is flagged hotspare, the raid card should automatically start the rebuild if a drive fails. You should never have to tell it to do that. If you had to tell it then it hardly qualifies as a hot spare. I was referring to the fact that softraid couldn't do that. Hot-spare devices work just fine (see below). What doesn't exists afaik are global hot spares. E.g. 7 disks, two 3 disk raid5 and one spare disk for whatever raid fails first. You would have to script that yourself. Ah. Bummer. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/28/2009 02:50 AM, Goswin von Brederlow wrote: [snip] The only argument I see for FC is a switched sorage network. As soon as you dedicate a storage box to one (or two) servers there is really no point in FC. Just use a SAS box with SATA disks inside. It is a) faster, b) simpler, c) works better and d) costs a fraction. The Tier 1 vendors can be touchy about certifying SATA SANs in certain environments, especially 24x7 DCs. That's why only our tier 3 (there is not tier 2...) storage is SATA. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. 1GBit is satturated by a single good disk already. 1GBit is a joke for fast storage. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Alex Samad a...@samad.com.au writes: On Wed, Feb 25, 2009 at 05:06:30PM -0500, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. true, depends on whos rule of thumb you use. I have seen places where mandate fc drives only in the data center - get very expensive when you want lots of disk space. The only argument I see for FC is a switched sorage network. As soon as you dedicate a storage box to one (or two) servers there is really no point in FC. Just use a SAS box with SATA disks inside. It is a) faster, b) simpler, c) works better and d) costs a fraction. And hey, we are talking big disks space for a single system here. Not sharing one 16TB raid box with 100 hosts. Also the disk space might not be need for feeding across the network, db aren't the only thing that chew through disk space. the op did specific enterprise, I was think very large enterprise, the sort of people who mandate scsi or sas only drives in their data centre They have way to much money and not enough brain. MfG Goswin PS: The I in RAID stands for inexpensive. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Ron Johnson ron.l.john...@cox.net writes: On 02/25/2009 03:48 PM, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Ah. The minicomputer tradition I come from (and thus how I organized my home PC) is to have a relatively small OS/swap disk and a separate data array. Of course, max device size always gets bigger, and smaller devices fall off the market... I'm aiming for a small SSD disk for the system and seperate data array that can be spun down most of the time. It doesn't take much with modern SATA drives to hit 2TB. Given we can get 1.5TB in a single drive, how many months before we can get 2TB in a single disk. Later this year. Waiting for them. 1.5TB disks don't mix so well with 1TB disks in raid. I don't want to split the disks into 0.5TB partitions and then raid over those. Anyone know how/if windows copes with 2TB disks? Does it understand GPT too? MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Ron Johnson ron.l.john...@cox.net writes: On 02/26/2009 02:51 PM, Alex Samad wrote: [snip] I have gone through a few cycles of changing the underlying drive sizes, ie a 3 disk raid5 made up of 3 x 500Gb and replacing in line with 3 x TB. pop 1 disk replace with 1 TB once it has settled you can do an online expansion. Not sure if you can do that on a HW raid. You used to not be able to. Not sure about modern controllers. Depends on the firmware and roughly speaking the price of your raid. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Fri, Feb 27, 2009 at 07:51:29AM +1100, Alex Samad wrote: my rule of thumb is to always have atleast 2 partitions on the first 2 drives (3 if I have them), for a raid1 /boot and a raid1 /. the rest of the space is put into a raid device then into lvm. That gets rid of the interesting tweaks. Even with software raid1, setting up reliable boot from either drive if one fails can be interesting, but it has gotten a lot better than it used to be. I asked about this in regards to grub2 the other day. The problem with software raid for me is that when I switch disks the drive order gets messed up every time. The first is that if the first disk fails hd1 becomes hd0 and so on. The other reason is that the onboard chips can't see SATA 2 disks. So if one of the onboard disks fails I have to move disks around to get another SATA 1 disk on the onboard port and free a port on the SATA 2 controler. And then the disk order is usualy scrambled up and grub fails. Now with grub2 you don't have to specify a boot device as (hd0,0) anymore but grub2 will suposedly find the right disk itself. This makes is really interesting for software raid. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Ron Johnson ron.l.john...@cox.net writes: On 02/27/2009 07:50 AM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 05:58:43PM -0600, Ron Johnson wrote: As would auto-replacement of bad drives by hot spares. Usually the firmware of a raid card does that itself. If a drive is flagged hotspare, the raid card should automatically start the rebuild if a drive fails. You should never have to tell it to do that. If you had to tell it then it hardly qualifies as a hot spare. I was referring to the fact that softraid couldn't do that. Hot-spare devices work just fine (see below). What doesn't exists afaik are global hot spares. E.g. 7 disks, two 3 disk raid5 and one spare disk for whatever raid fails first. You would have to script that yourself. MfG Goswin -- # mdadm --create -l1 -n2 /dev/md9 /dev/ram0 /dev/ram1 # mdadm --add /dev/md9 /dev/ram2 # cat /proc/mdstat Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] md9 : active raid1 ram2[2](S) ram1[1] ram0[0] 65472 blocks [2/2] [UU] # mdadm --fail /dev/md9 /dev/ram0 # cat /proc/mdstat Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] md9 : active raid1 ram2[0] ram1[1] ram0[2](F) 65472 blocks [2/2] [UU] And syslog shows: Feb 28 10:09:47 frosties mdadm[4078]: NewArray event detected on md device /dev/md9 Feb 28 10:10:00 frosties kernel: md: bindram2 Feb 28 10:10:52 frosties kernel: raid1: Disk failure on ram0, disabling device. Feb 28 10:10:52 frosties kernel: ^IOperation continuing on 1 devices Feb 28 10:10:52 frosties kernel: RAID1 conf printout: Feb 28 10:10:52 frosties kernel: --- wd:1 rd:2 Feb 28 10:10:52 frosties kernel: disk 0, wo:1, o:0, dev:ram0 Feb 28 10:10:52 frosties kernel: disk 1, wo:0, o:1, dev:ram1 Feb 28 10:10:52 frosties kernel: RAID1 conf printout: Feb 28 10:10:52 frosties kernel: --- wd:1 rd:2 Feb 28 10:10:52 frosties kernel: disk 1, wo:0, o:1, dev:ram1 Feb 28 10:10:52 frosties kernel: RAID1 conf printout: Feb 28 10:10:52 frosties kernel: --- wd:1 rd:2 Feb 28 10:10:52 frosties kernel: disk 0, wo:1, o:1, dev:ram2 Feb 28 10:10:52 frosties kernel: disk 1, wo:0, o:1, dev:ram1 Feb 28 10:10:52 frosties kernel: md: recovery of RAID array md9 Feb 28 10:10:52 frosties kernel: md: minimum _guaranteed_ speed: 1000 KB/sec/disk. Feb 28 10:10:52 frosties kernel: md: using maximum available idle IO bandwidth (but not more than 20 KB/sec) for recovery. Feb 28 10:10:52 frosties kernel: md: using 128k window, over a total of 65472 blocks. Feb 28 10:10:52 frosties kernel: md: md9: recovery done. Feb 28 10:10:52 frosties kernel: RAID1 conf printout: Feb 28 10:10:52 frosties kernel: --- wd:2 rd:2 Feb 28 10:10:52 frosties kernel: disk 0, wo:0, o:1, dev:ram2 Feb 28 10:10:52 frosties kernel: disk 1, wo:0, o:1, dev:ram1 -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Thu, Feb 26, 2009 at 05:42:43PM -0500, Douglas A. Tutty wrote: The comparison wasn't between having the raid controller or LVM present a reasonable size /, it was between a reasonable size / and a 2TB /. No one ever wanted a 2TB /. I just wanted / on a drive that was bigger than 2TB and hence couldn't use dos partition tables anymore. I only have a 10GB / :) Making a 10GB raid volume and a seperate raid volume for the rest just to be able to use dos partition tables for the / is just awkward. Not to repeat myself but a GPT with an entry for /boot in its fake MS-Dos table works just fine. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Goswin von Brederlow wrote: lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. 1GBit is satturated by a single good disk already. 1GBit is a joke for fast storage. Erm, not on anything other than a sequential read (and even then, I've never seen a single disk that would actually sustain that across it's whole capacity). Even raid-5s of significant numbers of disks aren't enormously fast, especially under multiple access. hdparm informs me that the SATA 28+2 spare raid-5 I have will read 170M a second. That would rapidly diminish under any sort of load. The only thing we've found that'll stand up to real multiuser load (like a mail spool) is raid-10, and enough spindles. We're beginning to see the requirement for 10GE on busy machines. -- ian MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Ian McDonald i...@st-andrews.ac.uk writes: Goswin von Brederlow wrote: lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. 1GBit is satturated by a single good disk already. 1GBit is a joke for fast storage. Erm, not on anything other than a sequential read (and even then, I've never seen a single disk that would actually sustain that across it's whole capacity). A cheap SATA disk with 7200rpm sustains 80MB/s sequential read/write on the outside and 40MB/s on the inside. An Seagate Cheetah 15K.6 is specified to up to 171MB/s and SAS disks are more uniform between outside and inside tracks. Even raid-5s of significant numbers of disks aren't enormously fast, especially under multiple access. hdparm informs me that the SATA 28+2 spare raid-5 I have will read 170M a second. That would rapidly diminish under any sort of load. For our Lustre filesystems we tested 16 SATA disks in an Infotrend SAS raid enclosure. As raid6 we still get 450 MiB/s sequential writing und 700MiB/s sequential reading. And that scales pretty well with more enclosures and more clients. In your case I would think the problem is your configuration. An 28 disk raid5 has a lot of stripes. That takes a lot of cache per stripe and a lot of cpu to calculate parity. Plus the chance of 2 disks failing before the spare disk can be synced mustbe HUGE. Have you ever thought about making multiple smaller raids? The only thing we've found that'll stand up to real multiuser load (like a mail spool) is raid-10, and enough spindles. Mail spool is like database access. Tons and tons of tiny read/write requests. The only thing that counts there is seek time. And the only raid level that improves seek time is raid1 (and the raid1 in raid10). We're beginning to see the requirement for 10GE on busy machines. Don't forget that you have overhead too. If you only have 1GBit to the storage then how is your server supposed to saturate the 1GBit to the ouside world? MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Goswin von Brederlow wrote: Ian McDonald i...@st-andrews.ac.uk writes: Goswin von Brederlow wrote: lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. 1GBit is satturated by a single good disk already. 1GBit is a joke for fast storage. Erm, not on anything other than a sequential read (and even then, I've never seen a single disk that would actually sustain that across it's whole capacity). A cheap SATA disk with 7200rpm sustains 80MB/s sequential read/write on the outside and 40MB/s on the inside. An Seagate Cheetah 15K.6 is specified to up to 171MB/s and SAS disks are more uniform between outside and inside tracks. My experience is that this sustained speed has quite a few lumps and bumps in it. I must admit, I thought we were talking about SATA disks, not recent SAS 15k's, and 40-80M/s is quite a way from 1 GBit. My WD Raptors only report around 75M/s. Even raid-5s of significant numbers of disks aren't enormously fast, especially under multiple access. hdparm informs me that the SATA 28+2 spare raid-5 I have will read 170M a second. That would rapidly diminish under any sort of load. For our Lustre filesystems we tested 16 SATA disks in an Infotrend SAS raid enclosure. As raid6 we still get 450 MiB/s sequential writing und 700MiB/s sequential reading. And that scales pretty well with more enclosures and more clients. In your case I would think the problem is your configuration. An 28 disk raid5 has a lot of stripes. That takes a lot of cache per stripe and a lot of cpu to calculate parity. Plus the chance of 2 disks failing before the spare disk can be synced mustbe HUGE. Have you ever thought about making multiple smaller raids? Of course. This performance isn't a problem for our requirement (given it's connected to 1GE), it's just illustrative. I'm not sure the risk of twin failure is that great, if you do calculations on MTBF's. Perhaps I ought to simulate a failure, and see how long it takes to rebuild :) We have a 56 disk + 4 spare Raid 10 on the production side of this setup, which is much much quicker :) (and still connected to 1GE, but can sustain multiple accesses well). The only thing we've found that'll stand up to real multiuser load (like a mail spool) is raid-10, and enough spindles. Mail spool is like database access. Tons and tons of tiny read/write requests. The only thing that counts there is seek time. And the only raid level that improves seek time is raid1 (and the raid1 in raid10). Indeed. We're beginning to see the requirement for 10GE on busy machines. Don't forget that you have overhead too. If you only have 1GBit to the storage then how is your server supposed to saturate the 1GBit to the ouside world? Who said I had 1G to the storage? The Storage is on 16x PCI-e, with 4x SAS connects to it :) Best Regards, -- ian -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 05:58:43PM -0600, Ron Johnson wrote: As would auto-replacement of bad drives by hot spares. Usually the firmware of a raid card does that itself. If a drive is flagged hotspare, the raid card should automatically start the rebuild if a drive fails. You should never have to tell it to do that. If you had to tell it then it hardly qualifies as a hot spare. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 06:06:07PM -0600, Ron Johnson wrote: Most DC managers have a bit more clue and good reasons than simply rules for rules' sake. Mainly logistics: if all the center's disks are SAS (or whatever other standard you choose) in only one or two vendor's SANs (or whatever other cabinet you choose), it makes the Operation staff's job a whole lot easier, thus helping to ensure greater uptime. Well people can choose to do it that way, at a huge increase in cost. I would have hoped the staff could manage a few types of disks without messing anything up. I guess it means having a few more types of spares around. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/27/2009 07:50 AM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 05:58:43PM -0600, Ron Johnson wrote: As would auto-replacement of bad drives by hot spares. Usually the firmware of a raid card does that itself. If a drive is flagged hotspare, the raid card should automatically start the rebuild if a drive fails. You should never have to tell it to do that. If you had to tell it then it hardly qualifies as a hot spare. I was referring to the fact that softraid couldn't do that. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Fri, Feb 27, 2009 at 11:49:29AM -0600, Ron Johnson wrote: I was referring to the fact that softraid couldn't do that. Are you sure? mdadm appears capable of managing spares automatically when such are setup for the raid. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/27/2009 02:25 PM, Lennart Sorensen wrote: On Fri, Feb 27, 2009 at 11:49:29AM -0600, Ron Johnson wrote: I was referring to the fact that softraid couldn't do that. Are you sure? No... mdadm appears capable of managing spares automatically when such are setup for the raid. In mdadm.conf? I'm really surprised (and pleased)! -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Fri, Feb 27, 2009 at 02:44:04PM -0600, Ron Johnson wrote: On 02/27/2009 02:25 PM, Lennart Sorensen wrote: On Fri, Feb 27, 2009 at 11:49:29AM -0600, Ron Johnson wrote: I was referring to the fact that softraid couldn't do that. Are you sure? No... mdadm appears capable of managing spares automatically when such are setup for the raid. In mdadm.conf? I'm really surprised (and pleased)! not sure about mdadm.conf. but definately when you define a raid set you can define a hot spare -- Ron Johnson, Jr. Jefferson LA USA -- The recession started upon my arrival. It could have been -- some say February, some say March, some speculate maybe earlier it started -- but nevertheless, it happened as we showed up here. The attacks on our country affected our economy. Corporate scandals affected the confidence of people and therefore affected the economy. My decision on Iraq, this kind of march to war, affected the economy. - George W. Bush 02/08/2004 on Meet the Press signature.asc Description: Digital signature
Re: big machines running Debian?
On Fri, Feb 27, 2009 at 02:44:04PM -0600, Ron Johnson wrote: In mdadm.conf? I'm really surprised (and pleased)! Probably in the monitoring mode. man mdadm talks about spare drives and spare groups and moving spares between raids and such. Sounds pretty likely to automatically use a spare assigned to an md raid though. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Fri, Feb 27, 2009 at 04:14:52PM -0500, Lennart Sorensen wrote: On Fri, Feb 27, 2009 at 02:44:04PM -0600, Ron Johnson wrote: In mdadm.conf? I'm really surprised (and pleased)! Probably in the monitoring mode. man mdadm talks about spare drives and spare groups and moving spares between raids and such. Sounds pretty likely to automatically use a spare assigned to an md raid though. I one (yesterday, to be precise) used mdadm to tell my RAID that one of its drives had failed, and it immediately started copying the other active member to the spare. -- hendrik -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
lsore...@csclub.uwaterloo.ca (Lennart Sorensen) writes: On Wed, Feb 25, 2009 at 04:07:54PM +0100, Goswin von Brederlow wrote: I think the limit is 1024 cores. Or was that fixed to allow more? I think people are working on that, but not too many machines need that yet. Most machines with that many cores are clusters and hence run multiple linux instances. As for ram that really is a cpu/architecture limit and you won't be able to find a motherboard that supports as much ram as the cpu(s) could handle. I can't remember if the current amd64/x86_64 architecture limit is 40bit or 44bit of physical memory space. Fairly decent chunk of ram either way, if you can fit it into the system in the first place. cat /proc/cpuinfo address sizes : 40 bits physical, 48 bits virtual which means 1TiB of ram. Good luck finding a board for that. More than 1TB on disk? Doh. 1TB fits on a single disk. Anything up to 16 TB is quite trivial. Beyond that you start to hit the limit on filesystem size with ext3 and have to use xfs or ext4 or something. Or you have to partition the space iinto 16TB chunks. Also 16 disks requires a big enough case or external storage. For that I would look into external enclosures with SAS connector. Don't use SCSI or FC. Well at 2TB you have to switch from DOS style partition tables to GPT, which requires the use of grub2 rather than lilo or grub, but works fine otherwise. Nah. Just needs /boot to be below 2TiB. The GPT has a fake MS-Dos table inside that grub can use and lilo only looks at the block addresses anyway. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 08:53:45PM -0600, Ron Johnson wrote: On 02/25/2009 07:22 PM, Douglas A. Tutty wrote: [snip] /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO Why is Read Ahead disabled on Logical Drive 1? My understanding is that read ahead in this case refers to the ability of the raid card to read ahead from one disk while a read is taking place on another disk. This only makes sense in a redundant raid level. LD1 is raid0, so there is no other disk from which to read ahead. If my understanding is off, I'd have to find it in the manual. Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
2009-02-26_14:21:54-0500 Douglas A. Tutty dtu...@vianet.ca: On Wed, Feb 25, 2009 at 08:53:45PM -0600, Ron Johnson wrote: On 02/25/2009 07:22 PM, Douglas A. Tutty wrote: [snip] /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO Why is Read Ahead disabled on Logical Drive 1? My understanding is that read ahead in this case refers to the ability of the raid card to read ahead from one disk while a read is taking place on another disk. This only makes sense in a redundant raid level. LD1 is raid0, so there is no other disk from which to read ahead. My understanding is that read ahead means the controller reads more data into memory than you asked for, expecting that the next bits you ask for will be immediately after the ones you just got. -- Ron Peterson Network Systems Manager Mount Holyoke College http://www.mtholyoke.edu/~rpeterso - I wish my computer would do what I want it to do - not what I tell it to do. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 05:37:12PM -0500, Douglas A. Tutty wrote: Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? Why should I do that? This way I can resize any LVM I want to any size I want. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 05:10:58PM -0600, Ron Johnson wrote: On 02/25/2009 04:37 PM, Douglas A. Tutty wrote: Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? I think it's because disk itself (which is what the boot loader sees) is .gt. 2TB. Well he is correct in that the raid controller could probably create multiple volumes from the raid to expose as devices to the OS. I find that overly complicated and less flexible though. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 11:41:03AM +1100, Alex Samad wrote: This begs the question why did you pick hardware raid over software raid You can boot from it no matter what (software raid can require interesting tweaks to the boot loader setup to make it work). Recovery can be transparent to the OS and be as simple as swapping out the drive that failed. You get nice hotswap bay LED control to show which drive has failed (I imagine software could do this too, but I have never seen that happen yet.) I have been a long supporter of software raid, but I find myself leaning towards a HP smart array 400 and using hardware raid (looking at 10 disks in raid6). My current thoughts are why should I have 10 channels (4 of them come from 1 pcix card) when I could have 1 channel to the smart array. there seem to be a few cciss utilities for me to track the array I am waying this up against the ability to easily manage the array and do upgrade and change disk and monitor the individual disks Some hardware raids have good support for monitoring under linux. Some do not. Having monitoring is quite important. The biggest advantage to software raid is that it is hardware independant. You can move all the disks to another controller type on another system, and linux's software raid will still work. Hardware raid setups are often very specific to one controller type so recovery from a controller failure can be tricky if you don't have access to spares. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 03:38:49PM -0500, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 11:41:03AM +1100, Alex Samad wrote: This begs the question why did you pick hardware raid over software raid You can boot from it no matter what (software raid can require interesting tweaks to the boot loader setup to make it work). my rule of thumb is to always have atleast 2 partitions on the first 2 drives (3 if I have them), for a raid1 /boot and a raid1 /. the rest of the space is put into a raid device then into lvm. That gets rid of the interesting tweaks. Recovery can be transparent to the OS and be as simple as swapping out the drive that failed. true You get nice hotswap bay LED control to show which drive has failed (I imagine software could do this too, but I have never seen that happen yet.) true I have been a long supporter of software raid, but I find myself leaning towards a HP smart array 400 and using hardware raid (looking at 10 disks in raid6). My current thoughts are why should I have 10 channels (4 of them come from 1 pcix card) when I could have 1 channel to the smart array. there seem to be a few cciss utilities for me to track the array I am waying this up against the ability to easily manage the array and do upgrade and change disk and monitor the individual disks Some hardware raids have good support for monitoring under linux. Some do not. Having monitoring is quite important. is that monitoring of the raid drives or the actual drives underneath, I like having smartctl to give me access to the actual drive health The biggest advantage to software raid is that it is hardware independant. You can move all the disks to another controller type on another system, and linux's software raid will still work. Hardware raid setups are often very specific to one controller type so recovery from a controller failure can be tricky if you don't have access to spares. I have gone through a few cycles of changing the underlying drive sizes, ie a 3 disk raid5 made up of 3 x 500Gb and replacing in line with 3 x TB. pop 1 disk replace with 1 TB once it has settled you can do an online expansion. Not sure if you can do that on a HW raid. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- I suspect that had my dad not been president, he'd be asking the same questions: How'd your meeting go with so-and-so? ... How did you feel when you stood up in front of the people for the State of the Union Address--state of the budget address, whatever you call it. - George W. Bush 03/09/2001 in an interview with the Washington Post signature.asc Description: Digital signature
Re: big machines running Debian?
On 02/26/2009 01:49 PM, Ron Peterson wrote: 2009-02-26_14:21:54-0500 Douglas A. Tutty dtu...@vianet.ca: On Wed, Feb 25, 2009 at 08:53:45PM -0600, Ron Johnson wrote: On 02/25/2009 07:22 PM, Douglas A. Tutty wrote: [snip] /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO Why is Read Ahead disabled on Logical Drive 1? My understanding is that read ahead in this case refers to the ability of the raid card to read ahead from one disk while a read is taking place on another disk. This only makes sense in a redundant raid level. LD1 is raid0, so there is no other disk from which to read ahead. My understanding is that read ahead means the controller reads more data into memory than you asked for, expecting that the next bits you ask for will be immediately after the ones you just got. That *is* the standard definition. Though there's nothing stopping Megaraid from being weird. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/26/2009 02:51 PM, Alex Samad wrote: [snip] I have gone through a few cycles of changing the underlying drive sizes, ie a 3 disk raid5 made up of 3 x 500Gb and replacing in line with 3 x TB. pop 1 disk replace with 1 TB once it has settled you can do an online expansion. Not sure if you can do that on a HW raid. You used to not be able to. Not sure about modern controllers. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 03:38:49PM -0500, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 11:41:03AM +1100, Alex Samad wrote: You get nice hotswap bay LED control to show which drive has failed (I imagine software could do this too, but I have never seen that happen yet.) Since the status of each physical drive is included in the /proc/megaraid/* then software that monitors the appropriate files will be able to tell you. This is one of the requirements for the software I'm designing. It will note when a drive isn't optimal and {syslog | email | wall | whatever}. Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 03:34:22PM -0500, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 05:37:12PM -0500, Douglas A. Tutty wrote: Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? Why should I do that? This way I can resize any LVM I want to any size I want. The comparison wasn't between having the raid controller or LVM present a reasonable size /, it was between a reasonable size / and a 2TB /. Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Fri, Feb 27, 2009 at 07:51:29AM +1100, Alex Samad wrote: my rule of thumb is to always have atleast 2 partitions on the first 2 drives (3 if I have them), for a raid1 /boot and a raid1 /. the rest of the space is put into a raid device then into lvm. That gets rid of the interesting tweaks. Even with software raid1, setting up reliable boot from either drive if one fails can be interesting, but it has gotten a lot better than it used to be. is that monitoring of the raid drives or the actual drives underneath, I like having smartctl to give me access to the actual drive health Well monitoring of raid health would be minimum. Getting more details would be nice. The biggest advantage to software raid is that it is hardware independant. You can move all the disks to another controller type on another system, and linux's software raid will still work. Hardware raid setups are often very specific to one controller type so recovery from a controller failure can be tricky if you don't have access to spares. I have gone through a few cycles of changing the underlying drive sizes, ie a 3 disk raid5 made up of 3 x 500Gb and replacing in line with 3 x TB. pop 1 disk replace with 1 TB once it has settled you can do an online expansion. Not sure if you can do that on a HW raid. Some hardware raids can do lots of things. Some can do no resizing at all. I have certainly used hardware raid cerads where adding a disk to a raid5 and expanding it was no problem. It just did it in the background, and when done you could reboot the system and the disk was suddenly bigger and software could do whatever it wanted to resize to the new larger disk. It also dealt with moving to larger disks in the raid by rebuilding one drive at a time, and then when all where replaced you could increase the raid to the size of the new disks. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 05:42:43PM -0500, Douglas A. Tutty wrote: The comparison wasn't between having the raid controller or LVM present a reasonable size /, it was between a reasonable size / and a 2TB /. No one ever wanted a 2TB /. I just wanted / on a drive that was bigger than 2TB and hence couldn't use dos partition tables anymore. I only have a 10GB / :) Making a 10GB raid volume and a seperate raid volume for the rest just to be able to use dos partition tables for the / is just awkward. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 11:36:20AM +1100, Alex Samad wrote: true, depends on whos rule of thumb you use. I have seen places where mandate fc drives only in the data center - get very expensive when you want lots of disk space. Also the disk space might not be need for feeding across the network, db aren't the only thing that chew through disk space. the op did specific enterprise, I was think very large enterprise, the sort of people who mandate scsi or sas only drives in their data centre Perhaps. I think some people make hard rules where in fact they would get a much better result by thinking instead for each case. Of course thinking can be hard, and it is much easier to just follow a hard rule so you don't get in trouble for making a decision. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/26/2009 05:49 PM, Lennart Sorensen wrote: On Fri, Feb 27, 2009 at 07:51:29AM +1100, Alex Samad wrote: my rule of thumb is to always have atleast 2 partitions on the first 2 drives (3 if I have them), for a raid1 /boot and a raid1 /. the rest of the space is put into a raid device then into lvm. That gets rid of the interesting tweaks. Even with software raid1, setting up reliable boot from either drive if one fails can be interesting, but it has gotten a lot better than it used to be. is that monitoring of the raid drives or the actual drives underneath, I like having smartctl to give me access to the actual drive health Well monitoring of raid health would be minimum. Getting more details would be nice. As would auto-replacement of bad drives by hot spares. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/26/2009 05:54 PM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 11:36:20AM +1100, Alex Samad wrote: true, depends on whos rule of thumb you use. I have seen places where mandate fc drives only in the data center - get very expensive when you want lots of disk space. Also the disk space might not be need for feeding across the network, db aren't the only thing that chew through disk space. the op did specific enterprise, I was think very large enterprise, the sort of people who mandate scsi or sas only drives in their data centre Perhaps. I think some people make hard rules where in fact they would get a much better result by thinking instead for each case. Of course thinking can be hard, and it is much easier to just follow a hard rule so you don't get in trouble for making a decision. Ehh. Most DC managers have a bit more clue and good reasons than simply rules for rules' sake. Mainly logistics: if all the center's disks are SAS (or whatever other standard you choose) in only one or two vendor's SANs (or whatever other cabinet you choose), it makes the Operation staff's job a whole lot easier, thus helping to ensure greater uptime. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 06:49:38PM -0500, Lennart Sorensen wrote: On Fri, Feb 27, 2009 at 07:51:29AM +1100, Alex Samad wrote: my rule of thumb is to always have atleast 2 partitions on the first 2 [snip] Some hardware raids can do lots of things. Some can do no resizing at all. I have certainly used hardware raid cerads where adding a disk to a raid5 and expanding it was no problem. It just did it in the background, and when done you could reboot the system and the disk was suddenly bigger and software could do whatever it wanted to resize to the new larger disk. It also dealt with moving to larger disks in the raid by rebuilding one drive at a time, and then when all where replaced you could increase the raid to the size of the new disks. Interesting, well I guess you get what you pay for (presuming the more you pay the better the card and the more features), I haven't seen any hardware based raid controllers that allow for increasing in size of the underlying disk's. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- Life is too short to be taken seriously. -- Oscar Wilde signature.asc Description: Digital signature
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 06:06:07PM -0600, Ron Johnson wrote: On 02/26/2009 05:54 PM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 11:36:20AM +1100, Alex Samad wrote: [snip] Perhaps. I think some people make hard rules where in fact they would get a much better result by thinking instead for each case. Of course thinking can be hard, and it is much easier to just follow a hard rule so you don't get in trouble for making a decision. Ehh. Most DC managers have a bit more clue and good reasons than simply rules for rules' sake. Mainly logistics: if all the center's disks are SAS (or whatever other standard you choose) in only one or two vendor's SANs (or whatever other cabinet you choose), it makes the Operation staff's job a whole lot easier, thus helping to ensure greater uptime. for large site, server owners would ask for space and not for disks, it would be upto to the storage guys to provide it and they like every one else like to KISS and thus usually standardise. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- Why don't you ever enter any CONTESTS, Marvin?? Don't you know your own ZIPCODE? signature.asc Description: Digital signature
Re: big machines running Debian?
On 02/26/2009 10:27 PM, Alex Samad wrote: On Thu, Feb 26, 2009 at 06:06:07PM -0600, Ron Johnson wrote: On 02/26/2009 05:54 PM, Lennart Sorensen wrote: On Thu, Feb 26, 2009 at 11:36:20AM +1100, Alex Samad wrote: [snip] Perhaps. I think some people make hard rules where in fact they would get a much better result by thinking instead for each case. Of course thinking can be hard, and it is much easier to just follow a hard rule so you don't get in trouble for making a decision. Ehh. Most DC managers have a bit more clue and good reasons than simply rules for rules' sake. Mainly logistics: if all the center's disks are SAS (or whatever other standard you choose) in only one or two vendor's SANs (or whatever other cabinet you choose), it makes the Operation staff's job a whole lot easier, thus helping to ensure greater uptime. for large site, server owners would ask for space and not for disks, it would be upto to the storage guys to provide it and they like every one else like to KISS and thus usually standardise. Very true. Also the server owners don't like to pay a lot, so there's negotiation and needs clarification... -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Umarzuki Mochlis umarz...@gmail.com writes: 2009/2/21 Igor Támara [[i...@tamarapatino.org]] Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. If Red Hat can support it, I don't think that there's any reason Debian couldn't. I think the limit is 1024 cores. Or was that fixed to allow more? As for ram that really is a cpu/architecture limit and you won't be able to find a motherboard that supports as much ram as the cpu(s) could handle. More than 1TB on disk? Doh. 1TB fits on a single disk. Anything up to 16 TB is quite trivial. Beyond that you start to hit the limit on filesystem size with ext3 and have to use xfs or ext4 or something. Or you have to partition the space iinto 16TB chunks. Also 16 disks requires a big enough case or external storage. For that I would look into external enclosures with SAS connector. Don't use SCSI or FC. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 04:07:54PM +0100, Goswin von Brederlow wrote: I think the limit is 1024 cores. Or was that fixed to allow more? I think people are working on that, but not too many machines need that yet. Most machines with that many cores are clusters and hence run multiple linux instances. As for ram that really is a cpu/architecture limit and you won't be able to find a motherboard that supports as much ram as the cpu(s) could handle. I can't remember if the current amd64/x86_64 architecture limit is 40bit or 44bit of physical memory space. Fairly decent chunk of ram either way, if you can fit it into the system in the first place. More than 1TB on disk? Doh. 1TB fits on a single disk. Anything up to 16 TB is quite trivial. Beyond that you start to hit the limit on filesystem size with ext3 and have to use xfs or ext4 or something. Or you have to partition the space iinto 16TB chunks. Also 16 disks requires a big enough case or external storage. For that I would look into external enclosures with SAS connector. Don't use SCSI or FC. Well at 2TB you have to switch from DOS style partition tables to GPT, which requires the use of grub2 rather than lilo or grub, but works fine otherwise. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, 25 Feb 2009, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 04:07:54PM +0100, Goswin von Brederlow wrote: More than 1TB on disk? Doh. 1TB fits on a single disk. Anything up to 16 TB is quite trivial. Beyond that you start to hit the limit on filesystem size with ext3 and have to use xfs or ext4 or something. Or you have to partition the space iinto 16TB chunks. Also 16 disks requires a big enough case or external storage. For that I would look into external enclosures with SAS connector. Don't use SCSI or FC. Well at 2TB you have to switch from DOS style partition tables to GPT, which requires the use of grub2 rather than lilo or grub, but works fine otherwise. Only if you want partitions, we usually don't for large data filesystems where the large filesystem sizes are relevant. As Goswin mentioned, you probably want to look at a different filesystem than ext3 for non-trivial fs sizes, not only due to limits but also perforamance. /Mattias Wadenstein -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 04:51:44PM +0100, Mattias Wadenstein wrote: Only if you want partitions, we usually don't for large data filesystems where the large filesystem sizes are relevant. If you have a seperate OS disk, then sure, partitions are not necesary, and even LVM and such have no need for partitions. As Goswin mentioned, you probably want to look at a different filesystem than ext3 for non-trivial fs sizes, not only due to limits but also perforamance. Certainly true. I am not having issues with ext3 on 2TB filesystems, but 16TB might be a different story. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/25/2009 09:14 AM, Lennart Sorensen wrote: [snip] Well at 2TB you have to switch from DOS style partition tables to GPT, which requires the use of grub2 rather than lilo or grub, but works fine otherwise. Who boots off of (or puts / on) a 2TB partition? -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. It doesn't take much with modern SATA drives to hit 2TB. Given we can get 1.5TB in a single drive, how many months before we can get 2TB in a single disk. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 04:07:54PM +0100, Goswin von Brederlow wrote: Umarzuki Mochlis umarz...@gmail.com writes: 2009/2/21 Igor Támara [[i...@tamarapatino.org]] Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. If Red Hat can support it, I don't think that there's any reason Debian couldn't. I think the limit is 1024 cores. Or was that fixed to allow more? As for ram that really is a cpu/architecture limit and you won't be able to find a motherboard that supports as much ram as the cpu(s) could handle. More than 1TB on disk? Doh. 1TB fits on a single disk. Anything up to most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB 16 TB is quite trivial. Beyond that you start to hit the limit on filesystem size with ext3 and have to use xfs or ext4 or something. Or you have to partition the space iinto 16TB chunks. Also 16 disks requires a big enough case or external storage. For that I would look into external enclosures with SAS connector. Don't use SCSI or FC. MfG Goswin -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- If the terriers and bariffs are torn down, this economy will grow. - George W. Bush 01/01/2000 signature.asc Description: Digital signature
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 08:54:11AM +1100, Alex Samad wrote: most enterprise site don;t use 1TB size disk, if you want performance you go spindles, there might be 8 disks (number pulled from the air - based on raid6 + spares) behind 1TB And if you want disk space and are serving across a 1Gbit ethernet link, you don't give a damn about spindles and go for cheap abundant storage, which means SATA. Not everyone is running a database server. Some people just have files. Raid5/6 of a few SATA drives can easily saturate 1Gbit. And for a very small fraction of the cost of SAS drives. -- Len Sorensen -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/25/2009 04:26 PM, Ian McDonald wrote: Ron Johnson wrote: On 02/25/2009 03:48 PM, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Ah. The minicomputer tradition I come from (and thus how I organized my home PC) is to have a relatively small OS/swap disk and a separate data array. Of course, max device size always gets bigger, and smaller devices fall off the market... It doesn't take much with modern SATA drives to hit 2TB. Given we can get 1.5TB in a single drive, how many months before we can get 2TB in a single disk. Later this year. Last month actually.. http://www.reghardware.co.uk/2009/01/27/review_internal_hard_drive_wd_caviar_green_2tb/ And at only a 15% premium to two 1TB drives... http://www.newegg.com/Product/Product.aspx?Item=N82E16822136337 http://www.newegg.com/Product/Product.aspx?Item=N82E16822136344 -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 04:48:30PM -0500, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/25/2009 04:37 PM, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 04:48:30PM -0500, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? I think it's because disk itself (which is what the boot loader sees) is .gt. 2TB. -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 05:10:58PM -0600, Ron Johnson wrote: On 02/25/2009 04:37 PM, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 04:48:30PM -0500, Lennart Sorensen wrote: On Wed, Feb 25, 2009 at 02:55:09PM -0600, Ron Johnson wrote: Who boots off of (or puts / on) a 2TB partition? Someone with a 4 drive raid5 on a hardware controller with 750GB SATA drives. Hence the only drive in the system is a 2.25TB device with partitions and everything on it. The root partition isn't very big, but it's on a drive that is bigger than 2TB and hence needs something other than a DOS partition table. Why wouldn't you configure the raid controller to give you a small logical drive (with whatever raid config you want) for the OS, and the larger logical drive for your data (or for LVM for everything except /)? I think it's because disk itself (which is what the boot loader sees) is .gt. 2TB. Not with my NetRaid card. It takes the physical disks and assembles them into virtual disks which appear to the OS as sd* of whatever size. Info on the underlying physical drives (and the virtual disks and the controller) show up under /proc/megaraid. Here's dmesg | grep -i scsi: SCSI subsystem initialized scsi0:Found MegaRAID controller at 0xf8814000, IRQ:177 scsi0 : LSI Logic MegaRAID F 254 commands 16 targs 4 chans 7 luns scsi0: scanning scsi channel 0 for logical drives. Type: Direct-Access ANSI SCSI revision: 02 Type: Direct-Access ANSI SCSI revision: 02 scsi0: scanning scsi channel 4 [P0] for physical devices. SCSI device sda: 5120 512-byte hdwr sectors (26214 MB) SCSI device sda: 5120 512-byte hdwr sectors (26214 MB) sym0: SCSI BUS has been reset. sym0: SCSI BUS mode change from SE to SE. scsi1 : sym-2.2.3 sym0: SCSI BUS has been reset. sd 0:0:0:0: Attached scsi disk sda SCSI device sdb: 39858176 512-byte hdwr sectors (20407 MB) SCSI device sdb: 39858176 512-byte hdwr sectors (20407 MB) sd 0:0:1:0: Attached scsi disk sdb The megaraid controller shows up as a scsi hba with two drives (sda, sdb) on it. In this case, sda is a raid1 array and sdb is a raid0 array; the OS knows nothing about this, however. Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Wed, Feb 25, 2009 at 07:08:13PM -0500, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 05:10:58PM -0600, Ron Johnson wrote: On 02/25/2009 04:37 PM, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 04:48:30PM -0500, Lennart Sorensen wrote: [snip] Not with my NetRaid card. It takes the physical disks and assembles them into virtual disks which appear to the OS as sd* of whatever size. This begs the question why did you pick hardware raid over software raid ? I have been a long supporter of software raid, but I find myself leaning towards a HP smart array 400 and using hardware raid (looking at 10 disks in raid6). My current thoughts are why should I have 10 channels (4 of them come from 1 pcix card) when I could have 1 channel to the smart array. there seem to be a few cciss utilities for me to track the array I am waying this up against the ability to easily manage the array and do upgrade and change disk and monitor the individual disks [snip] Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org -- Security is the essential roadblock to achieving the road map to peace. - George W. Bush 07/25/2003 Washington, DC signature.asc Description: Digital signature
Re: big machines running Debian?
On Thu, Feb 26, 2009 at 11:41:03AM +1100, Alex Samad wrote: On Wed, Feb 25, 2009 at 07:08:13PM -0500, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 05:10:58PM -0600, Ron Johnson wrote: On 02/25/2009 04:37 PM, Douglas A. Tutty wrote: On Wed, Feb 25, 2009 at 04:48:30PM -0500, Lennart Sorensen wrote: [snip] Not with my NetRaid card. It takes the physical disks and assembles them into virtual disks which appear to the OS as sd* of whatever size. This begs the question why did you pick hardware raid over software raid The HP NetServer LPr PII-450 boxes (4) I bought came with: two CPUs 1 GB ram two 72 GB SCSI drives (hot-swap) and the HP NetRaid-1si card. All for, IIRC, $65 CDN. I have been a long supporter of software raid, but I find myself leaning towards a HP smart array 400 and using hardware raid (looking at 10 disks in raid6). My current thoughts are why should I have 10 channels (4 of them come from 1 pcix card) when I could have 1 channel to the smart array. there seem to be a few cciss utilities for me to track the array All the status info ends up in /proc/megaraid. True, there aren't any utilities to monitor it. The card does have an alarm, although it isn't testable under the Linux Megaraid driver (it is under OpenBSD's driver). I'm starting work on a monitoring program. Actually, I'm using it as an exercise to refresh my structured analysis and design technique skills (they're over 20 years rusty [God, it is that long; I'm getting old]. I'll do it in Ada, modularized so that if the proc intefaces changes I only have to change that module. I am waying this up against the ability to easily manage the array and do upgrade and change disk and monitor the individual disks sure, you can't (under Linux) make new arrays on the card (requries booting into the bios) although there is supposed to be a dos program; I wonder if it can run under one of the dos emulators for linux. The individual disks can be monitored via the /proc interface. /proc/megaraid: hba0/ /proc/megaraid/hba0: battery-status config diskdrives-ch0 diskdrives-ch1 diskdrives-ch2 diskdrives-ch3 mailbox raiddrives-0-9 raiddrives-10-19 raiddrives-20-29 raiddrives-30-39 rebuild-rate stat /proc/megaraid/hba0/diskdrives-ch0 Channel: 0 Id: 0 State: Online. Vendor: HPModel: 36.4GB C 80-D94N Rev: D94N Type: Direct-Access ANSI SCSI revision: 02 Channel: 0 Id: 1 State: Online. Vendor: HPModel: 36.4GB C 80-D94N Rev: D94N Type: Direct-Access ANSI SCSI revision: 02 /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO This shows that both hard drives are Online. Both Raid drives are in an optimal state. I hope this helps your decision making. If you want to send me your ideas for the requriments for the monitoring software, I'll encorporate it in my system analysis. Thanks, Doug. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On 02/25/2009 07:22 PM, Douglas A. Tutty wrote: [snip] /proc/megaraid/hba0/raiddrives-0-9 Logical drive: 0:, state: optimal Span depth: 1, RAID level: 1, Stripe size: 64, Row size: 2 Read Policy: Adaptive, Write Policy: Write thru, Cache Policy: Cached IO Logical drive: 1:, state: optimal Span depth: 0, RAID level: 0, Stripe size:128, Row size: 0 Read Policy: No read ahead, Write Policy: Write thru, Cache Policy: Cached IO Why is Read Ahead disabled on Logical Drive 1? -- Ron Johnson, Jr. Jefferson LA USA The feeling of disgust at seeing a human female in a Relationship with a chimp male is Homininphobia, and you should be ashamed of yourself. -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
Hi, Dave On Saturday, 21.02.2009 at 08:00 -0500, Igor Támara wrote: Dave Dave Hi, at some Datacenter here on my country they only want the machines Dave to be installed with RHEL or Suse, every time I dig more into those Dave distros I fall in love more with Debian. This is why I'm asking about Dave machines that have many cores and lots of RAM and plenty of disk. Dave Dave We're running a Dell R905 server: four quad-core CPUs, 128GB RAM, with Dave an attached Dell Powervault storage system running off Dell's PERC 6/E Dave controller. Dave Dave This basically Just Works under Debian Etch (and, I suspect, Lenny too). Dave Thanks Peter and Dave, the report of the both machines I got is really important, thanks a lot for letting us know, that as one could suspect, Debian can be used in such environments. thank you all. Dave Dave. Dave Dave -- Dave Dave Ewart Dave da...@ceu.ox.ac.uk Dave Computing Manager, Cancer Epidemiology Unit Dave University of Oxford / Cancer Research UK Dave PGP: CC70 1883 BD92 E665 B840 118B 6E94 2CFD 694D E370 Dave Get key from http://www.ceu.ox.ac.uk/~davee/davee-ceu-ox-ac-uk.asc Dave N 51.7516, W 1.2152 -- Recomiendo Audacity para hacer edición de audio http://audacity.sourceforge.net signature.asc Description: Digital signature
Re: big machines running Debian?
Igor Támara wrote: Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. Good experiences with IBM blades, DS4200 SAN and Qlogic FC adapters. No Debian friendly SAN/FC multipath support available from IBM, but the debian package multipath-tools managed everything fine eventually. My educated guess is that the solution is by no means limited to IBM SANs. -Kyuu -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Tue, Feb 24, 2009 at 02:28:11PM -0500, Igor Támara wrote: Hi, Dave On Saturday, 21.02.2009 at 08:00 -0500, Igor Támara wrote: Dave Dave Hi, at some Datacenter here on my country they only want the machines Dave to be installed with RHEL or Suse, every time I dig more into those Dave distros I fall in love more with Debian. This is why I'm asking about Dave machines that have many cores and lots of RAM and plenty of disk. Dave Dave We're running a Dell R905 server: four quad-core CPUs, 128GB RAM, with Dave an attached Dell Powervault storage system running off Dell's PERC 6/E Dave controller. Dave Dave This basically Just Works under Debian Etch (and, I suspect, Lenny too). Dave Thanks Peter and Dave, the report of the both machines I got is really important, thanks a lot for letting us know, that as one could suspect, Debian can be used in such environments. I have a HP DL785 - 8 socket quad core amd with 64G of memory, installed of a debian installer snapshot. I have it attached to a HP EVA8000 (8T) + HP XP (4T) , both with multipath tool managing the san. It boot of local disk smart arrays. I had some issue with their psp pack (support pack), but that was because they were waiting for debian 5 to release. thank you all. Dave Dave. Dave Dave -- Dave Dave Ewart Dave da...@ceu.ox.ac.uk Dave Computing Manager, Cancer Epidemiology Unit Dave University of Oxford / Cancer Research UK Dave PGP: CC70 1883 BD92 E665 B840 118B 6E94 2CFD 694D E370 Dave Get key from http://www.ceu.ox.ac.uk/~davee/davee-ceu-ox-ac-uk.asc Dave N 51.7516, W 1.2152 -- Recomiendo Audacity para hacer edición de audio http://audacity.sourceforge.net -- Bill wrote a book at Yale. I read one. - George W. Bush 10/19/2000 New York City, NY on William F. Buckley, Al Smith Dinner signature.asc Description: Digital signature
Re: big machines running Debian?
On Saturday, 21.02.2009 at 08:00 -0500, Igor Támara wrote: Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. We're running a Dell R905 server: four quad-core CPUs, 128GB RAM, with an attached Dell Powervault storage system running off Dell's PERC 6/E controller. This basically Just Works under Debian Etch (and, I suspect, Lenny too). Dave. -- Dave Ewart da...@ceu.ox.ac.uk Computing Manager, Cancer Epidemiology Unit University of Oxford / Cancer Research UK PGP: CC70 1883 BD92 E665 B840 118B 6E94 2CFD 694D E370 Get key from http://www.ceu.ox.ac.uk/~davee/davee-ceu-ox-ac-uk.asc N 51.7516, W 1.2152 signature.asc Description: Digital signature
big machines running Debian?
Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. Is there a place where one can post the machines to make some feel of trusting for others? I'm using Debian from about 2000 and had the opportunity of use sparc, powerpc, x86 and AMD64 ports, and ever had to go back with another distro, I'm really happy with Debian, so I want to use it as many places as possible. Thanks in advacne for any information. -- Recomiendo Imágenes de OpenClipart http://www.openclipart.org signature.asc Description: Digital signature
Re: big machines running Debian?
2009/2/21 Igor Támara i...@tamarapatino.org Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. If Red Hat can support it, I don't think that there's any reason Debian couldn't. Is there a place where one can post the machines to make some feel of trusting for others? I'm using Debian from about 2000 and had the opportunity of use sparc, powerpc, x86 and AMD64 ports, and ever had to go back with another distro, I'm really happy with Debian, so I want to use it as many places as possible. Thanks in advacne for any information. -- Recomiendo Imágenes de OpenClipart http://www.openclipart.org -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFJn/rwtV4JcpE0AlYRAkuvAJ9Lk6TLC1WmyjcJ2m1qpiBC/jkDegCg70+5 yDLpMTq6odSmddpuYWYHkPQ= =Md58 -END PGP SIGNATURE- -- Regards, Umarzuki Mochlis http://gameornot.net
Re: big machines running Debian?
I don't know about their size specs, but both linode and slicehost let you set up your own distro, mostly coloc though. Nuno Magalhães LU#484677 -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
On Sat, Feb 21, 2009 at 9:00 PM, Igor Támara i...@tamarapatino.org wrote: Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Hi Igor, nice to meet you. It is very good to know that you also like Debian. I think this depends on the kernel so it really shouldn't matter which distribution you use. We have NPTL now for threads thus I things should be just fine. Best regards, Nelson.- -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: big machines running Debian?
From: Igor T?mara i...@tamarapatino.org Date: Sat, Feb 21, 2009 at 08:00:32AM -0500 Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. My hosting provider, dreamhost, runs Debian. Individual machines may not meet your specs, but they do have a lot of them. Jurriaan -- What does ELF stand for (in respect to Linux?) ELF is the first rock group that Ronnie James Dio performed with back in the early 1970's. In constrast, a.out is a misspelling of the French word for the month of August. What the two have in common is beyond me, but Linux users seem to use the two words together. seen on c.o.l.misc -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
RE: big machines running Debian?
Here at Vulcan We run some 2x4 processor 32GB RAM 2TB Arrays via fibre channel without a problem with Lenny Debian Peter Yorke From: Igor Támara [i...@tamarapatino.org] Sent: Saturday, February 21, 2009 5:00 AM To: debian-amd64@lists.debian.org Subject: big machines running Debian? Hi, at some Datacenter here on my country they only want the machines to be installed with RHEL or Suse, every time I dig more into those distros I fall in love more with Debian. This is why I'm asking about machines that have many cores and lots of RAM and plenty of disk. Here (at my country) big means more than 4x4 cores , more than 16Gb of RAM, and more than 1Tb on disk, excluding clusters, also SAN are good to know about. Is there a place where one can post the machines to make some feel of trusting for others? I'm using Debian from about 2000 and had the opportunity of use sparc, powerpc, x86 and AMD64 ports, and ever had to go back with another distro, I'm really happy with Debian, so I want to use it as many places as possible. Thanks in advacne for any information. -- Recomiendo Imágenes de OpenClipart http://www.openclipart.org -- To UNSUBSCRIBE, email to debian-amd64-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org