Upgrading Snap 4500

[blue68f100]

Thanks for the debug page, looks like it got clipped at a column length of 35.

Nothing jumped out at me, that may help you.

[webboy]

Nicely done it seems that you can also type linux/unix commands it apears that the debug page is really a sh command prompt

[kuoman]

I am new to SNAP server. I am in the process of getting four 250GB HD (replacing the four 120GB currently on the machine). It is running OS 3.2.019 with RAID5 setup.
When I get these four 250GB, what is the upgrading method? I heard, swap one HD at a time and let the new HD sync (hours?) then do the same for the other three HDs.

So which drive do I start first? from drive 1, 2, 3 and then 4 or from drive 4, 3, 2 and 1? Some people mention about losing the OS, will swapping one at a time remove the OS on the system?

[blue68f100]

Boot with drive 1. The hot swap drive 2. Thenpower down and move drive 2 to 1. Reboot, then hotswap the rest of the drives.

[Phoenix32]

Break your RAID 5 first or the OS will make the new drive match the old drive size.

[kuoman]

Originally Posted by Phoenix32

Break your RAID 5 first or the OS will make the new drive match the old drive size.

By "break" you mean delete RAID 5 right? Which RAID set should I use for upgading the HD?

blue68f100 replied in another post (I won't do doble post, so I will reply on this thread). I don't care about the data, since I have it backed up already. I will do a factory rest, disable CA Antivuris, disable Snapshot then shutdown the system. Remove drive 2, 3 and 4 then power it back on. Insert the new HD to slot 2.
After the system sync drive 1 (old) with drive 2 (new). I power it down and take out drive 1 and 2. Insert new drive (was drive 2) into drive 1 slot.
Power the system up (only slot 1 has drive), I put in new drive in slot 2, 3 and 4 (can I put all 3 at the same time or I have to do one at a time after the drive is synced?)

[Rote8]

The Snap 4500 uses an "invisible N-Way RAID" for the OS in the first 100 MB of each hard drive.
Replacing the drives, one at a time will work, but the data RAID set will still be the original size.
To use the full size of the new drives the original data RAIDset must be deleted and recreated.

Because of this I advise the following method:
Backup the data on the RAID.
Remove the original RAIDset.
Replace 3 drives. (keep the originals untouched, in case something goes wrong)
Go have lunch while the OS RAIDset syncs the N-Way raid.
Once the OS RAIDset is repaired, replace the 4th drive.
Create a new RAIDset using the full drive size.

[blue68f100]

By "break" you mean delete RAID 5 right?

YES, You must break/delete the raid array other wise it will revolk recovery console to diaster recovery.

Replace 3 drives. (keep the originals untouched, in case something goes wrong)
Go have lunch while the OS RAIDset syncs the N-Way raid.

WHY, it only takes a few minutes for this process.

[Phoenix32]

David, why do people insist on making this harder or more complicated than it is?

I have done this several hundred times with various GOS revision, hard disk brands, and hard disk sizes without a problem, so listen up.

This is a simple step by step that assumes you do not want to keep your current data (or have backed it up elsewhere). It also assumes you have a working GOS on the unit.

I do break the current data raid before I start, but not always and you don't have to because once you remove the drives as I describe here, its gone anywise. I just do it so I don't get errors in the log telling me it blew up.

1. Remove drives 2 thru 4. Leave drive 1 (or a seed drive) in slot 1.
2. Put one of your new drives in slot 2. (no drives in slot 3 or 4)
3. Boot the unit.

a) I tell people to wait about 15 minutes from here.
b) It takes 5 to 9 minutes to boot the unit depending on your unit and configuration etc. It only takes a couple minutes after full boot for the OS to get copied over from slot 1 to slot 2, but give it 5 to 10 minutes from full boot up to be sure. No big deal.

4. Shut the unit down.
5. Remove the old drive from slot 1.
6. Put your new drive from slot 2 that just got the OS installed into slot 1.
7. Put your other 3 fresh clean new drives into slots 2 thru 4.
8. Boot the unit.
9. Wait 10 minutes from full boot up for the OS to copy to the 3 new drives.

c) At this point, I usually do a hardware reset and a factory default reset and erase the logs etc to clean things up, but not required.

10. Your done.


A child could do it.


Edit Note: This also assumes you were using RAID 5 (by far the most common). If not, then you will need to break the array first.

[bcole]

blue68f100,

I was able get my self out of some jams with advice you've posted here
and was wondering if you have any thoughts on this one

attempting to update my 4500 after another HD failure an repeated reloading
failures, I figured i go ahead and break the raid which i needed to do anyways
in order to install larger HDs now the system is only loading grub Am I sol? is there a way to use grub to relink back to where the os is? I've looked at the drive in with koppix and it appears the guardian stuff is still there.

I do have a full install gsu for my version 4.0.228

all recomendations welcome

Thanks for listening

[Phoenix32]

Originally Posted by bcole

blue68f100,

I was able get my self out of some jams with advice you've posted here
and was wondering if you have any thoughts on this one

What am I, chopped liver?

Sounds like part of the OS is maybe corrupt.

Answers are;
No
Yes

[bcole]

Originally Posted by Phoenix32

What am I, chopped liver?

Sorry, didn’t mean to insult you or anyone else
You could possibly be Filet Mignon unless you’re a vegetarian in which case I maybe insulting you again
Never the less I merely pointed to the individual who helped me out before. Help me out here and I’ll
Look upon you as whatever top grade food of your choice, precious metals or gems.
Are we good yet?
I can grovel even more……. In the mean time please help
Thanks again

[Phoenix32]

Find out what IP it is grabbing (like from your router etc). Watch the lights on the front and see if it is going into recovery mode (the light will flash at a normal rate, but will be flashing between yellow and amber). If so, cool, log in via the browser GUI and reload your GSU file.

if no...

Since your data is backed up and the raid broken anywise, pull the drives and try to boot the unit with each drive, one at a time alone, in slot 1, and see if the unit will boot with any of them.

[blue68f100]

Originally Posted by bcole

blue68f100,

I was able get my self out of some jams with advice you've posted here
and was wondering if you have any thoughts on this one

attempting to update my 4500 after another HD failure an repeated reloading
failures, I figured i go ahead and break the raid which i needed to do anyways
in order to install larger HDs now the system is only loading grub Am I sol? is there a way to use grub to relink back to where the os is? I've looked at the drive in with koppix and it appears the guardian stuff is still there.

I do have a full install gsu for my version 4.0.228

all recomendations welcome

Thanks for listening

Sounds like your may had reference to your older hd's some where. The reason I do not want any raid, AV, snap shots or any software ref to HD's. I'm surprised that it does not launch the recovery console.

I would install one of your original HD and do the OS transfer to the new HD's.

[bcole]

I don't believe it's really booting all the way
all drives act the same (they are the originals not new)
with terminal hooked up I only get prompted with grub
here's what I'm doing to get it where it looks as if it's booted

grub> root (hd0,1)
Filesystem type is xfs, partition type 0x83

grub>kernel /boot/vmlinuz-gos-up root=hda2
[Linux-bzimage, setup=0xa00, size=0x141ae2]

grub> boot

Uncompressing kernel... OK, booting the kernel.
************************************************** ********
************************************************** ********
This device cannot be managed via video/kbd/mouse interface.
The video display output is now disabled
You may access the managment functions from the web your browser.
Thank You,
- Snap Appliance, Inc -
************************************************** ********
************************************************** ********

and there she sits only power & network led no heartbeat
no 'SNAP406572' on net work (default name)
no 'SNAP-1' (original assigned name)
no ip's given out from dhcp (was grabbing 100.100.11.12 before which is first available ip)
not useing 10.10.10.10
not using oringinal assigned ip 100.100.3.3


am I missing something here?

Thanks

[blue68f100]

Like you said it's looking for something that's not there.

From grub I would try to launch the recovery console. But I do not know what it's name actually is. It should be located on the first partition. Then do a clean install.

Have you replaced any of ram or any thing hardware?

[bcole]

not sure why but the hda1 (1st partition) was empty it should have all the *up and *smp files from /boot
on hda2 (2nd partition) + the /grub dir with a grub.conf file. there is script in /install_tools called mkboot.sh inside of it I could see that it creates the grub.conf from the type of system your running copied the portion for my system and created the /grub/grub.conf file. after that when I rebooted my hyperterminal was allowing me to see what was going on
no more hangs. able to see that I found the proper ip and could connect the recovery panel reinstall the os. Guardian also does not seem to want to propagate to drives which have a portion of the same ver OS
I’m assuming all you need to do is smoke the kernels from partition 1&2 and it would treat it as if it was
new and replicate itself.
New related issue
I read that you should be able in debug run “co de format 10000 /reinit” to have snap reinit the drive to it’s full capacity but Guardian OS isn’t recognizing the command is that a SnapOS specific command
Or am I still missing something?

Thanks again

FYI
1st partition contains
a----- 2776780 30-Nov-109 14:36 Guardian-initrd.smp
a----- 2766523 30-Nov-109 14:36 Guardian-initrd.up
a----- 407025 30-Nov-109 14:36 System.map-gos-smp
a----- 385298 30-Nov-109 14:36 System.map-gos-up
a----- 19649 30-Nov-109 14:36 config-gos-smp
a----- 19627 30-Nov-109 14:36 config-gos-up
-d---- 0 30-Nov-109 14:36 grub
-d---- 0 30-Nov-109 14:36 lost+found
a----- 1320674 30-Nov-109 14:36 vmlinuz-gos-smp
a----- 1250189 30-Nov-109 14:36 vmlinuz-gos-up

./grub:
total 130
-d---- 0 30-Nov-109 14:36 .
-d---- 0 30-Nov-109 14:36 ..
a----- 709 30-Nov-109 14:36 grub.conf
a----- 512 30-Nov-109 14:36 stage1
a----- 131008 30-Nov-109 14:36 stage2

[Phoenix32]

#1 Did you do as I told you the first time?

Originally Posted by Phoenix32

Find out what IP it is grabbing (like from your router etc). Watch the lights on the front and see if it is going into recovery mode (the light will flash at a normal rate, but will be flashing between yellow and amber). If so, cool, log in via the browser GUI and reload your GSU file.

if no...

Since your data is backed up and the raid broken anywise, pull the drives and try to boot the unit with each drive, one at a time alone, in slot 1, and see if the unit will boot with any of them.

#2 If yes, what was the results?

#3 If no, why should I post anything further?

[Phoenix32]

That's what I thought.

[bcole]

it wasn't until I redid the first partition (with knoppix) that the system responded,

Originally Posted by bcole

I don't believe it's really booting all the way
"all drives act the same "
"and there she sits only power & network led no heartbeat"

(middle led out = system not up)

I'm thinking the reason it didn't boot properly from grub was the kernel command
I used “kernel /boot/vmlinuz-gos-up root=hda2 “
root(hd0,1) <-- tells grub to look in 2nd partition since my 1st partition was empty
-----------------------actual guardian grub.conf points to (hd0,0) because the kernel should be there
-----------------------next line is more consistent with kernel statement in the grub.conf file
-----------------------with the exception of /boot which tells which dir the kernel (vmlinuz-gos-up)
-----------------------is located on the partition specified.

“kernel /boot/vmlinuz-gos-up root=/dev/ram ramdisk=16384 console=ttys0,115200n8 rw uppermem 16384”

------------------------I think lacking the console statement was preventing anything being output to the
------------------------hyperterminal conneted to the serial port.there is also a serial and terminal
------------------------statements at the beginig of grub.conf
------------------------ All I got was the same banner output to the vga. There should be 3 options to boot in a terminal setup
-------------------------1st option is normal 2nd being alternate boot 3rd which would boot to recovery console
If anyone knows for sure please verify, I'm only speculating. Strange thing is after system was repaired the /boot dir on partition disapeared
Right now once my system is back up and synced I’ll try to take a day and try to bring it back up with one of the drives where the OS is broke.
The system is up now and resyncing the new drives which had no problem copying the OS or recognizing their size.

Many Thanks To All,

Originally Posted by bcole

FYI
1st partition contains
a----- 2776780 30-Nov-109 14:36 Guardian-initrd.smp
a----- 2766523 30-Nov-109 14:36 Guardian-initrd.up
a----- 407025 30-Nov-109 14:36 System.map-gos-smp
a----- 385298 30-Nov-109 14:36 System.map-gos-up
a----- 19649 30-Nov-109 14:36 config-gos-smp
a----- 19627 30-Nov-109 14:36 config-gos-up
-d---- 0 30-Nov-109 14:36 grub
-d---- 0 30-Nov-109 14:36 lost+found
a----- 1320674 30-Nov-109 14:36 vmlinuz-gos-smp
a----- 1250189 30-Nov-109 14:36 vmlinuz-gos-up

./grub:
total 130
-d---- 0 30-Nov-109 14:36 .
-d---- 0 30-Nov-109 14:36 ..
a----- 709 30-Nov-109 14:36 grub.conf
a----- 512 30-Nov-109 14:36 stage1
a----- 131008 30-Nov-109 14:36 stage2

also
grub.conf (for snap4500 single processor)
================================================== ========
serial --unit=0 --speed=115200
terminal --timeout=0 serial

default 0
fallback 5
timeout 10

title Opus (default boot)
root (hd0,0)
kernel /vmlinuz-gos-up root=/dev/ram ramdisk=16384 console=ttyS0,115200n8 rw
uppermem 16384
initrd /Guardian-initrd.up

title Opus-alt (use alt kernel from /boot in root filesystem)
root (hd0,1)
kernel /boot/bzImage.alt root=/dev/ram ramdisk=16384 console=ttyS0,115200n8 rw ACCEPTROOT=1
uppermem 16384
initrd /boot/Guardian-initrd.alt

title tserver (start tserver installation only)
root (hd0,0)
kernel /vmlinuz-gos-up root=/dev/ram ramdisk=16384 console=ttyS0,115200n8 rw NOPIVOT=true
uppermem 16384
initrd /Guardian-initrd.up
================================================== ========