I am setting up a Raid5 and torture testing it. I added two eSata ports to my machine. When a drive is installed in that eSata port and the machine then booted up the device name (e.g. /dev/sdc) is inserted in the middle of my Raid devices. And that is just one example of how the device names can change.I did a search on 'static device names' but I saw nothing directly related to Raid. What I did see were suggestions to create udev rules based on UUID. But that was for single disks, not Raid, where each drive/partition in the raid array appears to have the same UUID.I'm surprised this does not come up in the various Raid howtos because it is impossible to keep a Raid array intact without solving this problem unless the machine is never touched thereafter.
I have just bought 4 1TB drives and set up a Software Raid level 5. Using Disk Utility tool I have created a GPT partition table and now when I want to create a partition, I get:
Error creating partition: helper exited with exit code 1: In part_add_partition: device_file=/dev/md0, start=0, size=3000610848768, type= Entering MS-DOS parser (offset=0, size=3000610848768) MSDOS_MAGIC found found partition type 0xee => protective MBR for GPT
[Code].....
it does say nothing about creating an partition on the /dev/md0, although Gnome Disk Utility allows to do that - if I just run mkfs.xsf /dev/md0 - it works fine, yet Disk Utility tells me that disk has not been partitioned, see image:
Something weird happened last night and my raid5 failed. I am trying to re activate it and see if my data is dead or what. When I run mdadm -Asv /dev/md0 I get
Code: mdadm: looking for devices for /dev/md0 mdadm: cannot open device /dev/dm-1: Device or resource busy mdadm: /dev/dm-1 has wrong uuid. mdadm: cannot open device /dev/dm-0: Device or resource busy mdadm: /dev/dm-0 has wrong uuid. mdadm: cannot open device /dev/sde2: Device or resource busy mdadm: /dev/sde2 has wrong uuid. mdadm: cannot open device /dev/sde1: Device or resource busy mdadm: /dev/sde1 has wrong uuid. mdadm: cannot open device /dev/sde: Device or resource busy mdadm: /dev/sde has wrong uuid. mdadm: cannot open device /dev/sdd: Device or resource busy mdadm: /dev/sdd has wrong uuid. mdadm: cannot open device /dev/sdc: Device or resource busy mdadm: /dev/sdc has wrong uuid. mdadm: cannot open device /dev/sdb: Device or resource busy mdadm: /dev/sdb has wrong uuid. mdadm: cannot open device /dev/sda: Device or resource busy mdadm: /dev/sda has wrong uuid.
I'm having trouble with Ubuntu 10.10 and stable device names. When I installed Ubuntu, the root drive was the only one in the machine; it obviously got /dev/sda.
After the base installation, I installed three additional 2TB drives to make RAID-5 array. Ubuntu renamed the root drive to /dev/sdd. While annoying I lived with it.
After creating a single partition set to "Linux raid autodetect" on each drive, I created the RAID-5 array:
Code:
All was going well until a reboot. When rebooting Ubuntu decided to make the root drive /dev/sda this time and now mdadm --detail /dev/md0 reports:
Code:
How to fix the array and make the device names stable?
This is one of those questions where I highly suspect that the answer will be something like ". . .just stinks bein' you, kid."My setup:I have Ubuntu 10.04 x64 (desktop) installed on a computer with a bunch of SATA ports, including one e-SATA on the backBecause of the way the hardware works with these ports, the *external* port is the first SATA port, a second SATA port inside becomes #2, and a cluster of four ports (that can be made into a (ahem!) RAID array become ports 3 through 6.
All ports are configured as SATA (PATA emulation)My boot drive is located on SATA-2 (the inside connector), a four disk "MD" raid array is located on SATA 3-6, and I have an external HD enclosure that I plug into the e-SATA connection.
With all six drives plugged in, they get ordered like this: /dev/sda - External drive /dev/sdb - Internal operating system drive (bootable)
I built a Raid5 volume with 3 SATA II hard disk drives. Further I have a system disk conected through IDE. During the first setup the IDE disk becomes sda, the SATA II disk sd[bcd] respectively. Now, sometimes the device names change after reboot - why ever... E.g. one of the raid5 disk become sda and so I got an error message during the boot procedure regarding the raid set. Curious, when the system is up and I stop and restart the Raid5 volume it comes up and runs fine. Because I'm currently at work I can't post any more detailed config files at the moment.
I am building a custom RedHat+<our-software> installer iso for our own appliance. I am using Red Hat 5.4. The appliances has two on-board eth interfaces. On the back panel of the appliance, these ports are marked 1 and 2. When I install RH, I find the device names assigned such as eth0/eth1 are arbitrary. I understand this is to be expected with kernels 2.6+. Most of our customers connect their eth cables to the port marked 1 on the back and assume they should configure eth0 to make the device reachable. However, sometimes port 1 gets assigned "eth1". This is not a blocking issue, but its going to confuse our customers and we wanted to make it easier on them.
From reading online discussion boards, I know HOW to switch the assignment of the eth names. However, what I am do not know is whether I need to switch them at all. So I have two questions 1) Is there anyway for me to tell which eth mac corresponds to which port on the back? Since they are soldered on the motherboard and not movable, I would think there would be some way to figure out that x mac address corresponds to the upper port (marked '1' etc). 2) Is there a way to tell this by running a linux command? We need to do this automatically so I need to be able to figure it out at install time from the kickstart post-install or similar.
I have a server wich is connected to an iSCSI storage and gets harddisks from this storage. Sometimes I have to add new disks to this server. Everytime I add a disk and make an /etc/init.d/iscsi restart on the server the disks don't have the same device name as before the iscsi restart.
It should be possible to gave the disks persistent names using udev rules. Now I tried to create different rules in "/etc/udev/rules.d/99-static-iscsi-names.rules" e.g.
# /dev/sdc KERNEL=="sd*", BUS=="scsi", PROGRAM=="/sbin/scsi_id -g -s %p",RESULT=="360a98000503355344c4a576864467877" NAME=="sdc%n" In "/etc/rc.local" I added "/sbin/start_udev" and in "/etc/scsi_id.config" I added the line "vendor="NETAPP",model="LUN",options=-g"
Ubuntu was a good choice for free server O/S. I am currently downloading it. My ISP wants to bind my static IP to a MAC address of the device that will be making my connection for me. Should my server be behind my router or in front of it? I kinda wanted it to be the access point and be able to monitor the usage from all the terminals in my house. We have 4 other PC's, a laptop and a PS3. My server is a Acer Altos G530 with 3.2 Xeon Processor, 4 gb of ECC DDR,
My os is debian 5.0, netcard is RTL8168D/8111D. I can't use the Static ip, if I setup static ip, output "not found network device", but the network driver have loaded and if I use DHCP mode everything is ok. do you understand my english? What my next steps should be?
I'm using an external USB drive to perform system backups. The script I'm using has a mount command - mount /dev/sdcx /systembak, it works fine until the external drive is mounted into a different USB port which causes the device path to change and the script needs to be modified to reflect that change. Looking in the /dev/disk/ directory I see ./by-id ./by-label ./by-path ./by-uuid. I've been able to mount the disk using one of the device pointers in those directories; are any of those device pointers static or do they change every time the USB disk is plugged into a different USB port.
Any basic description of how linux assigns drive letters? I understand that a drive letter assignment is not static. If I add a drive between /dev/sda and /dev/sdb, my /dev/sdb will become /dev/sdc and the new drive will become /dev/sdb. I have a hot swap tray and have come into some unexpected behavior. I removed /dev/sde from the hot swap tray and then loaded another drive into this same tray. When I mounted the new drive with options in fstab, it wouldn't mount because the new drive was /dev/sdf, not /dev/sde. Apparently, linux is looking at the id of the drive in addition to it's place in the BIOS chain.
My fstab entry is: /dev/sde /backups auto noauto,rw,noexec,async,user 0 0 I was avoiding using UUIDs in the fstab so that new HDDs would not have to be "registered" in the fstab prior to use. Is there a way to tell linux (or fstab) whatever drive is plugged into SATA channel X mount to /mountpoint?
I have had a software raid running on OpenSuse / SUSE 10 & 11 for some time and I have about 8TB on it. I always run close to 100% and once in a while get there but always manage to delete a bunch and clear out about 30-100GB.
I am using a 3rd party kernel driver that does not support udev properly. When I was using wheezy I placed the required device files in /lib/udev/devices.
The udev in jessie does not appear to support this. Is there any way to have udev create these device files or will I have to create then using a script at boot-up?
I have no drive failures but just need to recreate a raid5 set as the next free MD disk number. Originally I built a temp OS of debian on a single drive and had 4x2TB drives in a raid5 software array (MD0) this worked fine and allowed me to move all data to it, and remove our old fileserver. I have now pulled out the 4 x 2TB Raid 5 drives and created a new OS on two new 80GB drives, partioned as follows,
MD0 is now 250mb Raid1 as /boot MD1 is 4GB Raid1 Swap MD2 is 76GB Raid1 as /
If I turn off and push back in the 4x2TB drives I cannot see a MD3. I presume I would need to create a MD3 from these 4 drives but I dont want to mess things up as its live data. So im here asking for help, or a bit of hand holding to get it done right.
PS - Its a Debian Lenny 5.0.3 Raid1 fresh install replacing a Debian Lenny 5.0.3 on a single disk.
In my understanding, the way /proc/scsi/scsi gets populated, /proc/paritions also gets populated in the same fashion. i.e. the description for first entry of /proc/scsi/scsi can be seen in the first entry of /proc/partitions and same for rest.
So, With this assumption, in my project, I used to relate first entry of /proc/scsi/scsi with first entry of /proc/partitions to get its total size and same for all entries.
But, I observed some differences in following scenario, where
1) The first 4 entries in /proc/scsi/scsi are SAN luns attached to my system and for which the actual device names in /dev/ are sda,sdb,sdc and sdd.
2) The last 4 entries are the internal HDDs on same system. In /dev/, their respective device names are sde,sdf,sdg & sdh.
(Output attached at end of the thread)
But in /proc/partitions, the device order is different.
You can see their respective sizes in /proc/partition output as well.
So, my question is, in this particular scenario, I can't relate the first entry of /proc/scsi/scsi with first entry of /proc/partition. i.e. scsi0:00:00:00 is not /dev/sde, because it is actually /dev/sda.
It seems that my assumption is wrong in this scenario.
Is there any way or mechanism to figure out actual device name for an entry in /proc/scsi/scsi in /dev/ directory?
How can my application should relate /proc/scsi/scsi entries with their respective device names and sizes?
I'm having some problems on the line with ***. This is from a OpenEntidade.java class file and I'm trying to update EmprestadorView's variables.
EmprestadorView.java public class EmprestadorView extends FrameView { ... OpenEntidade.java ----------------- public class OpenEntidade extends javax.swing.JFrame {
I got this message on Friday from just one domain. uote:mailsrv.forthnet.gr #<mailsrv.forthnet.gr #5.5.0 smtp; 554 5.5.0 Your message was considered to be spam by the FORTHnet Antispamming Policy and was not delivered to the recipient. The following spam tests returned positive for this message:FORGED_RCVD_HELO,RCVD_IN_BRBL. For further information visitWe are not a spamming community but it seems we have a statice IP address that has a Reverse lookup to "myipaddress.static.lyse.net" and not my email domain. Would setting a cname mail.mydomain.no -> myipaddress.static.lyse.net cure this problem or are there more tricks to be performedOnce I have cured the FORGED_RCVD_HELO I can move to getting the IP removed from BARACUDA.
I have created a 9TB raid from 4 1.5TB drives and 3 2TB drives (1.6 and .4 partition). I thought it would be a 9TB partition, and Ubuntu says it is a 9TB partition except when looking at what drive space is left.Nautilus' Properties and System Monitor both say that the raid is 4.1TB with 1TB free but Disk Utility and Nautilus say it is a 9TB RAID. very odd. I have tried checking the raid and the xfs file system. no errors.here is from watch cat /proc/mdstat
3 new 1.5TB HD. 1 used 1.5TB hd with 980MB of data. I want to set up a raid 5 with a hot spare. I have music, pictures, videos, and movies (About 2.8TB worth). I have had a mismatch of drives previously, 250GB, 2 320GB, 500GB, 2 1TB and now a 1.5TB all with data. I have removed the one 250 and 2 320s and put the data on the 1.5TB that is currently installed.
What I would like to do is create a raid5 with the three new 1.5TB HD's, copy the data over from the currently installed 1.5TB and then grow or add that drive as a hot spare. Or just add it and then add another 1.5TB down the road as a hot spare don't know for sure.
In addition since I have 2 1 TB drives, I could add 2 more (Good deals on 1 TB drives right now) and have a total of 4 1TB drives. Could I have 2 raid5's (4-1TB's and 4-1.5TB's)in two separate arrays? I really do not know if that makes sense or not but here comes LVM. I am tired of managing my HD space and since i have multiple folders (Movies, music, pictures, videos) and within the movies folder I have R, G & PG folders for the ratings of the movies. (Pwd protect the R so the kids can't get to it) So with LVM installed with the Raid5 I should be able to create my folders and just keep adding data and not worry about moving folders around when I grow the storage by adding new drives. Is that correct? Maybe someone could point me to a how to.
Also, if I create 2 arrays (And I need to know so I can order the 2 additional 1TB drives), then I could put all the music, G and PG content on the one array and all the R and spicy stuff on the other and password protect it.
I am trying to build a new array after adjusting TLER on my disks, which permanently changed some of the drives sizes. I am not sure if the following inconsistencies are related to the newly mismatched drive sizes.
Using:
Code: mdadm --create --auto=md --verbose --chunk=64 --level=5 --raid-devices=4 /dev/md1 /dev/sdd /dev/sde /dev/sdf /dev/sdg Nets me (build-time was two full days):
[Code]....
On a side note, since I'm recreating my array from scratch, I was wondering if anyone here knows of any optimized settings I could use. I've got 3Tb of data to transfer, so lots of test material.
These are Western Digital First Generation 2TB Green Drives (WD20EADS-00R6B0) with WDidle3 fix applied & TLER=ON. These are pre Advanced Format (aka not 4K).
I used a raid 5 of 5*1TB via Kernel raid on md0. I then created a luks out of md0 -> /dev/mapper/md0. I formatted ext3 this container. That worked fine. Now I baught another Harddrive and grew the raid to now 6 Devices. The Raid is now running und bigger than before - great. The luksOpen still works BUT I can't mount the ext3 on it. It seems like the luks Volume also grew, it's now 1TB bigger without doing anything. The Problem: I can't mount the luks anymore, mapping it was no big deal so.. Always tells me that I should specify the filesystem type:
Code: root@ubuntu:/home/ubuntu# mount /dev/mapper/md0 /mnt/tmp mount: Sie missen den Dateisystemtyp angeben
If I do is then it continues to give me errors: Code: root@ubuntu:/home/ubuntu# mount /dev/mapper/md0 /mnt/tmp -t ext3 mount: wrong fs type, bad option, bad superblock on /dev/mapper/md0, missing codepage or helper program, or other error Manchmal liefert das Syslog wertvolle Informationen - versuchen Sie dmesg | tail oder so
It seems as if the luks system doesnt know that the luks container isn't using the whole drive.
I recently installed a new home backup server with Ubuntu 9.10 x86_64 using the alternate CD. I used the CD's installer to partition my disk and created a software RAID 5 array on 4 disks with no spares. The root file system is located outside the raid array.
At first the array performed nicely but as it started to fill up, the io performance dropped significantly to the point where I get a transfer rate of 1-2MB/s when writing!
Created my own file server/nas, but get stuck in a problem after couple of months. I have a server with 4x 1,5tb disks, all connected to sata ports and 1 40gb ata133 disk running ubuntu 9.10 x64 amd. I've created a raid5 array using mdadm. It all worked great for couple of months but lately the raid5 array is degraded. disk sdd1 is faulting every few days. I have checked the drive but it is fine. If I re-add the disk and wait for 6 hours my raid5 array is all fine again, but after a few shutdowns, it is degraded.
my mdadm detail:
Quote:
root@ubuntu: sudo mdadm --detail /dev/md0 /dev/md0: Version : 00.90 Creation Time : Mon Dec 14 13:00:43 2009 Raid Level : raid5
I have ubuntu server 10.04 on a server with 2.8ghz 1gb ddr2 with the os on a 2gb cf card attached to the IDE channel and a software raid5 with 4 x 750gb drives. On a samba share using these drives I am only getting around 5 MB/s connected via wireless N at 216mbps and my router and server both having gigabit ports. Is a raid 5 supposed to be that slow? I was seeing speeds of anywhere from 20-50MB/s from other people and am just wondering what i am doing wrong to be so far below that.
I recently went shopping at newegg for components for a new home server to fulfill multiple needs--chief amongst them to replace an aging old Athlon box which has been running a 4TB software raid for years now So I ordered an ASUS P7P55D-E LX mobo, 8GB DDR2, a core i5 quad-core, and 6 WD "green" 2TB drives (I'm more concerned with capacity than speed). I got it all assembled, using a 500GB 2.5" drive as the boot / host drive, a blu-ray reader, and splitting the 2TB drives amongst the 2 controllers (4 on the remaining ports of the x6 SATA controller and 2 on the x2 SATA-III controller, just not needing the SATA III capabilities). I plan to use Software RAID5 rather than the fakeraid that might be supported by the chipset. This affords maximum portability for the future.
At any rate, I installed 10.10 64-bit (Desktop), let it update, then installed mdadm and set about configuring the RAID5. I decided to use the nifty graphical tool. Last time I did this on my old Fedora box, it was all CLI.All seemed to be going well, and I was letting it fully recover / synchronize before setting up a filesystem (planning on ext4 this time). I kept an eye on it with cat /proc/mdstat.
After 20 hrs (averaging 25MB/sec on the sync...seems a bit slow, but who knows?)...it finishes and the RAID is immediately degraded with one volume (/dev/sdc1) marked as faulty and another volume marked as a spare. I replaced the SATA cables with brand new ones as long ago I ran into an issue like this due to a bad cable. Before trying again I ran the extensive SMART test and a read/write benchmark on each drive. They all checked out fine. So I started over and 20hrs later, same thing.So I shut down and removed the offending drive and started with only 5 drives. Another 19hrs go by and this one turns up /dev/sde1 faulty and another drive marked as a spare. That drive was fine before, and is on a different controller from the one that it was finding as faulty before.
Now I am suspicious that there is a larger underlying issue and that the drives are not bad at all. Should I have just gone with a 32-bit + PAE install? Does the 64-bit have known issues with the software RAID? Should I have just built it using the CLI? Looking online, it seems people are having success with the GUI tool. I don't think it is the GUI tool. Checking with mdadm, it seems everything is being configured right.
