Snapserver 4400 and 4500 questions

[[AfZ]PiMp J]

I've FINALLY gotten my hands on a 4500, that i've been waiting on for months now as phoenix knows

Snapserver 4400 Questions:

Question 1) Is the Cold BOOT OS Copying from drive 1 to other drives done via hardware or software?

Question 2) Does it only recognize / copy the GOS, or any OS?

1 of my 4400's had a hard drive DIE and even though it was in raid5, seems to have lost the data, so i figure now is the time to upgrade the drives, and while i'm at it install a new OS. Unfortunately i can't afford a new GOS (and the version i have is 3.2 i believe).

I've found this version to do most everything i want BUT be able to install an updated proftpd that allows for secure encryped connections and this is next to impossible due to missing libraries and/or compilers (i've outlined that issue in a previous post)

the OTHER 4400 has the same version of the OS but it has been suffering from the dreaded "no web page loading, but all my fileshares work and i can ssh into it" which seems to have the only remedy of re-installing the OS. I don't have the original CDs, nor do i have current space on the network to backup the data so i can break the RAID on it and pull the drive 1 copy trick.....which is why i'm upgrading the drives on the 4400.

Question 3) Since there is no monitor output from the 4400s, how WOULD someone install another OS on it, if they couldn't install it on a 4500 and use the drive copy trick mentioned above? I don't recall running across any posts on this.

This leaves the 4500

Question 1) Does anyone know of any issues with Red Hat 5 not being able to control the fans on this unit? (or even the 4400s for that matter?) I know of a post here by someone who installed FedoraCore 6 and their last post mentioned they couldn't get it to quiet down.

There's probably a question 2 for the 4500 but i'm tired of typing :P

thanks all

P.S. I may keep one of the 4400s running GOS and sell it to a friend. in this case i'll have some spare drives if they can be used by anyone here.

If he doesn't want it, i could see parting with the drives and the sleds, and i'd keep the rest for spare parts if i don't upgrade it

[blue68f100]

Question 1, The GOS cold boot copy to new HD is a software function.

Question 2, This is a feature built into the GOS.

The hardware is different between the 4400 and 4500. So installing on 4500 and moving to 4400 may or may not work.

I do know MS home server will install as well as most any Linux OS. But there are some flags that need set for it to boot.

On the 4400 you need to install a video card in the expansion slot, would help out on your installs of NON GOS software.

[Phoenix32]

If you were to see the later versions of GOS, you would not be wanting to switch. IMO, there is no better OS for NAS than the GOS by leaps and bounds (and the majority of the enterprise industry agrees with me). You really need to look at the later version GOS before you make such a drastic move.

For othrs who have the later versions of GOS, please chime in here so he can make an informed decision.

[[AfZ]PiMp J]

Well my quest got sidetracked anyway, as it appears red hat linux refuses to install itself on a software raid5. the boot partition can seemingly only be on a raid 1 setup at best, which would cut my disks in half and make raid5 undoable. I'm still gonna keep looking for a way around it though

[blue68f100]

The normal setup is that the GOS is on a 10gig partition in raid 1 ( all drives). With the remaining for users data.

[Phoenix32]

David, there are MANY more partitions than that. But, it doesn't matter in this particular case anymore, he found a solutions. Me.

[blue68f100]

Yes, but I was just pointing out that the GOS was in a raid1 on all HD's.

Glad you can help him out.

[[AfZ]PiMp J]

and thanks to you pointing that out, it clicked in my head that i was able to play with partition sizes and setup both a software raid 1 and software raid 5 on the same set of 4 drives, so i've got red hat running on the 4500 temporarily while i try and restore data from a 4400 before i part with it.

it just means i can now play around and try to figure out how to quiet the bitch down (or the hairdryer as you called it in another post)

side note: i've not much experience with a software raid setup, is it dependent on the OS surviving? for example in my situation if the raid 1 broke and destroyed the OS or even if the OS just became corrupt and unrecoverable, the data on the raid5 would no longer be salvagable right?

if the answer to that is yes, then how does the data on the software raid5 in GOS survive when the os itself is reinstalled?....or am i just assuming it does, when in reality the data doesn't survive?

@phoenix, i tried booting my 4400 with the bad drive last night, and it did boot, but the web page would not show up (previously it had) and i could not ssh in (connection refused) but i could try to access the unc path and get prompted for a login and password. i thought i correctly recalled both the root and admin default passwords but they didn't seem to let me in,

@all: is there another combo i could try to attempt to access the shares, or is windows access to the shares turned off by default on the OS so that no matter what i'm locked out?

[blue68f100]

Let me try to explain it. On a GOS hardware the OS is installed first. It allocates all of the needed partitions needed for the OS. This is then copied in most part across all the remaining HDs (raid1). When you create a Raid 5, it's a partition on each HD to make up the set. As long as you do not do a clean install the dada is still there. It's just a matter of accessing it. Which in most part can be done by re-storing the GOS, either from a recovery set or re-installing the OS through the recovery console. Now if you do what is a clean install it will wipe your HD clean. Depending on what GOS version was installed, a newer version may not have enough room to install. So the original version should be installed to backup the data before wiping every thing clean for the newer version.

I did some extensive testing when I got my 4500. I was pulling the power in the middle of large data transfers (writes) to try to crash the unit. As long as the watchdog was active, snapshots and diaster recovery active it was always able to recover. But soon as you took the diaster recovery away you were asking for trouble. In most cases if the Main HD boot failed you need to allow plenty of time for watchdog to attempt the next HD. This can be a long time considering the std boot takes anywhere from 5-10 min depending on model. One thing you can do is setup the null modem cable and your terminal program. This is active during the boot process so you can see where the problems are accouring.

If you must have the data, I would contact Snaptech. If only the OS was damaged preventing if from booting, extraction of the data will be quick and easy. And should not cost an arm and a leg.

One thing I do and Phoenix too is run SpinRite on ALL of our HD's. This at times will repair a HD and it will boot up. It also builds the bad sector table so the SMART does not have to do it during a transfer if a bad sector is found. This way will keep the HD's (RAID5) in sync.

You can SSH into the shell using a terminal program or putty on MS Windows if the gui is not working. You do NOT want Native Windows near the Snaps HDs. Another words if you can not connect via SMB, FTP or putty do not go any further with windows. It has a bad problem of trying to fix the HD so it can see it, which in turn destroys the data. If you want to snoop use a Linux flavor and terminal program. None of the partition slices are labeled so they do not make any sense. And most all the drives will have different names. You will see alot of symbolic links (shortcuts) so finding the right partition or folder can be time consuming.

Phoenix32 can help you hardware wise but is limited somewhat when it comes to the OS. But he's had several years now to catch up on the OS side.

Good luck.

[Phoenix32]

Originally Posted by blue68f100

Phoenix32 can help you hardware wise but is limited somewhat when it comes to the OS. But he's had several years now to catch up on the OS side.

I am a hardware guy (real edu-mi-cated and everything), and most likely the only one here. Thus my software strengths lay within the scopes of where the software and hardware cross. i.e. I know the formats used (the partition setups and so forth) and the various ins and outs of OS installations and issues. When it comes to bringing the hardware to life with an OS, that is my department (getting it installed and finding workarounds etc etc). Those are the areas I excel in. But, once the hardware and OS are up and running, then the software guys like David take over for the uses of that OS and other software. Make sence?

[[AfZ]PiMp J]

Originally Posted by blue68f100

I did some extensive testing when I got my 4500. I was pulling the power in the middle of large data transfers (writes) to try to crash the unit. As long as the watchdog was active, snapshots and diaster recovery active it was always able to recover. But soon as you took the diaster recovery away you were asking for trouble. In most cases if the Main HD boot failed you need to allow plenty of time for watchdog to attempt the next HD. This can be a long time considering the std boot takes anywhere from 5-10 min depending on model. One thing you can do is setup the null modem cable and your terminal program. This is active during the boot process so you can see where the problems are accouring.

If you must have the data, I would contact Snaptech. If only the OS was damaged preventing if from booting, extraction of the data will be quick and easy. And should not cost an arm and a leg.

One thing I do and Phoenix too is run SpinRite on ALL of our HD's. This at times will repair a HD and it will boot up. It also builds the bad sector table so the SMART does not have to do it during a transfer if a bad sector is found. This way will keep the HD's (RAID5) in sync.

You can SSH into the shell using a terminal program or putty on MS Windows if the gui is not working.

I may try the null modem cable to see what it spews if anything, but do know that i CANT ssh at this time (i do use putty), as i get a connection refused. I also CANT get into the gui. The data's not so important that i'd need to spend money to fix it, but i would like to muck around with it myself.

I've heard this spinrite mentioned a few times, do you have a favorite place to download it from or purchase it from if it's not a freebie? Also i assume it's windows based so i'd have to take the dead drive out of the snapserver to have spinrite scan it, or is that incorrect?

[blue68f100]

The only place to get Spinrite it is at www.GRC.com

It is Assembly language and runs at the controller level, so it does not care what the HD OS or format is.

You will need to pull the HD, Spinrite will be the boot disk, CD or Floppy.

[[AfZ]PiMp J]

I was actually talking about the host operating system to run the software, which thanks to the link shows me windows will be fine.

hope to have more extensive playtime this weekend

[[AfZ]PiMp J]

Ok so i've gotten my snap4400 to boot properly and allow me into the web page administration as well as putty to it, however it still thinks my raid5 is failed. Is there a command or two i can throw at it to have it force a rebuild in an attempt to save anything? I know in hardware raid setups you used to be able to recreate a stripe ontop of what had failed and that would sometimes let you recover your data, but this being software and raid5 as opposed to raid1 and linux based on top of that, thought i'd ask here and google while waiting

sorry lenn, i had hoped to ship this thing out to you on monday but i've got to try what i can before i do, in the interest of saving me multiple drive swaps and time with the 4400 i still have. i'll keep ya updated though

*edit*: I've found a few tools in the /install_tools directory, along the lines of mdctl and mkraid, so i'll keep looking around and update unless anyone here has experience with those tools and a suggested recovery method for muh raid5