logs archiveIRC Archive / Freenode / #centos / 2015 / September / 6 / 1
mock
hello. i am having some trouble managing my raid-5 array. i had a drive fail, but now there appear to be two missing drives from the array, and it will not start.
i need some help more interactive than google for this. is there a channel or someone who can assist?
DiscordianUK
You wear a Fedora cloak, you know better than to behave like this
Same tools apply in centos , mdadm etc
mock
DiscordianUK: don't judge. i'm a mdadm noob.
i've been running scans and assembles and details to do this. i am not making progress.
TrevorH
the mdadm mailing list is the definitive source of help
mock
TrevorH: i was hoping to not have to spend several days or weeks going back and forth to address this.
TrevorH
when it comes to your data, it's better to be cautious
mock
that's true, so the answer to my question is a "no"/"none" then?
DiscordianUK
Google finds this : http://www.tecmint.com/create-raid-5-in-linux/
Thermi
mock: Check if the two drives that are missing are faulty
And if they are indeed still usasble
         

mock
Thermi: one is merely removed (/dev/sda1). the other is most likely the faulty one, which is why i'm in this mess to begin with.
Thermi
Then add sda1 back, get the raid in degraded mode
mock
i cannot mdadm --add /dev/md0 /dev/sda1 while the array is mounted, but if i mdadm --stop /dev/md0, there command doesn't work because no /dev/md0 device found.
Thermi
*sigh* really?
First step now
do binary backups of the drive
of /all/ of them.
/NOW/
And /number them/
mock
not possible. they are all 2TB drives and i have no other storage device big enough for them.
Thermi
Then you go buy them. Do you have a backup of the data, that was on the RAID?
mock
this was my definitive backup solution... :\
Thermi
Are you fu*king kidding me?
centbot
Please keep language family-friendly while in this channel if you could be so kind. Thank you, we appreciate it.
mock
some, most of it, in various places
Thermi
You /must/ /always/ have a backup. A RAID is not a backup
TrevorH
how can you even have the array started if it's missing 2 drives?
Thermi
TrevorH: he doesn't.
DiscordianUK
He doesn't have a backup
Thermi
mock: Get sda1 back. Now.
TrevorH
" i cannot mdadm --add /dev/md0 /dev/sda1 while the array is mounted"
mock
Thermi: i have plans to make sure it's not the definite solution going forward, but it involves "cloud" as the backup backup
Thermi
Come back when you found it.
DiscordianUK
I sense dataloss will happen
Thermi
Cloud is not a backup.
mock: Do you have sda1?
         

mock
i can mdadm --examine /dev/sda1 with good results, but i cannot add while md0 is running
TrevorH
how can it be running?
DiscordianUK
A backup is only as good as the last time as it was test restored
Thermi
mock: You do not 'add' a device while running
'add' is for growing arrays
You do not want to grow the array
You want to 'assemble' it.
mock
ok
Thermi
With exactly ONE missing drive
When you assemble the array, you 'must' specify the drives in the /correct/ order
TrevorH
pastebin the output of mdadm --examine on all drives
Thermi
Emphasis is on *correct*
Lexically ordered is *not* the correct order
Show the configuration of mdadm
mock: What keeps you from assembling the array now?
TrevorH
he says that he's lost 2 drives but at the same time he says he has it running. One of those two things is not true
mock
TrevorH: so, the system drops me into a root shell upon boot because the raid problem. i'm limited in what i can network...if any.
Thermi
mock: Okay. Since you lost *two* drives while the system was online, your RAID is broken now
Did that happen?
TrevorH
use a rescue CD
Thermi
Did you loose one or *two* drives while the RAID was online?
A RAID5 can only survive the loss of one drive
If you removed one drive while the raid was offline, that's not a problem.
Put it back in.
Assemble the raid and tell mdadm what drive is missing now
mock
just one. i think one of the drives was added physically but not to the array.
Thermi
mock: What? Added 'physically but not to the array'`
?
When was it removed?
Was it ever in the array?
How many drives does the array have?
TrevorH
mock: what CentOS version is this?
mock
the assemble process won't work while mounted. when i mdadm --stop then --assemble, it tells me: /dev/md/0 assembled from 3 drives, not enough to start the array.
Thermi
mock: Show me your mdadm config and what drives you have.
mock
centos 6.6
Thermi
mock
i am working to get the information...
Haris
hello all
does lighttpd 1.4.37 come with or for 6.7 ?
mock
i have found the import parts of my data on a separate drive. if there is a way to salvage the array, i would like to try that. but if it is deemed a lost cause, i can explore the whole starting over process.
Haris
nope. comes with epel
TrevorH
epel6 has 1.4.36
mock: no one can help without data so we're waiting for the pastebin
mock
TrevorH: i know. i'm still getting that...
DiscordianUK
how long does running mdadm take?
TrevorH
I think it's getting the data off the system that is the problem if it won't boot properly
DiscordianUK
ahhh
TrevorH
mock: is the system installed on this array?
DiscordianUK
I wish my NAS ran CentOS
Thermi
mock: Read up on the link I last sent you
DiscordianUK
I do regular backups
mock
TrevorH: no, the system is installed on a separate drive. the /var directory is on the raid, though. a situation i will deal with in future installs.
Thermi: i will
Thermi
mock: Do not touch the raid unless you read the complete page
TrevorH
I'd also reckon that not much would need to be created on the underlying /var dir to make your system minimally bootable if you commented the existing /var fstab entry
Haris
hello all
how to downgrade a pkg, when I have the rpms' in the current folder, but not in the yum repos'
TrevorH
yum downgrade
Haris
that looks at the repo (epel). and fail subsequently
do I need to remove the current installed, and manually install the rpms of older version ?
« prev 1 2 3 4 next »