Xserver Fibre Channel with tape backup issues?

Is this correct, that OS X 10.4.9 and 10.4.10 has a fibre channel issue with external tape drives (as well as SCSI drives)???
And this is to be fixed in 10.4.11? Do we have a time frame when we will see 10.4.11 to fix this major snafu?
I'm going to decide on a different server (in the meantime) and do net backups if this is the case. But what about others that run backup off the server, attached to an Xserve RAID?
BTW, I contacted BRU support with query on their software...I give kudos to them. Their support is spot on!
Message was edited by: Eddie Ski

Retrospect is just hideously, egregiously out of date and I don't think EMC gives a d*mn, at all.
Even pre-Intel, it will happily LOCK up your server while failing to properly time out when trying to talk to a tape library. It's well-known, Retro + Tape Library = h*ll on earth. And the fault lies squarely with Retro because something like Bru and other more expensive software does not have the same issue.
So let's just make sure -not- to judge "SCSI" isues in OS X based on the ungodly-awful behaviour of Retrospect. Might as well do automobile performance tests using only amateur drivers ; )

Similar Messages

Help with tape backup

Hello,
I am trying to configure my database to use a tape backup device. I have Veritas Backup Exec running a Dell tape backup, when trying to test from Enterprise Manager I get the following error:
RMAN>
connected to target database: OHIO (DBID=1527591012)
using target database control file instead of recovery catalog
RMAN>
echo set on
RMAN> run {
2> allocate channel oem_sbt_backup type 'sbt_tape' format '%U';
3> backup as BACKUPSET current controlfile tag '07062006093228';
4> restore controlfile validate from tag '07062006093228';
5> release channel oem_sbt_backup;
6> }
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03009: failure of allocate command on oem_sbt_backup channel at 07/06/2006 09:32:33
ORA-19554: error allocating device, device type: SBT_TAPE, device name:
ORA-27211: Failed to load Media Management Library
RMAN> allocate channel for maintenance type 'sbt_tape' ;
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03009: failure of allocate command on ORA_MAINT_SBT_TAPE_1 channel at 07/06/2006 09:32:33
ORA-19554: error allocating device, device type: SBT_TAPE, device name:
ORA-27211: Failed to load Media Management Library
RMAN> delete noprompt backuppiece tag '07062006093228';
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of delete command at 07/06/2006 09:32:33
ORA-19554: error allocating device, device type: SBT_TAPE, device name:
ORA-27211: Failed to load Media Management Library
RMAN> exit;
I'm guessing I missed some kind of configuration step somewhere, but I am unsure what. Any advice would be greatly appreciated.
Regards,
Jordan

I would say OSB is really good product and performance wise also it is great. Though they haven't launched OSB for all of the platforms yet like AIX, 64bit Linux but all these are suppose to come by the end of Sep. I have been participated in beta testing of this product, it was great and now we are waiting for AIX and 64-bit linux releases so that we can start using them as primary backup tool.
It can take the file system backup too without any problem.
Daljit Singh

I want to upgrade from OS X 10.5.8 to Snow Leopard. When trying to back-up to a WD 1 TB external drive, the back-up process stalls at about 4 GB. Grateful for any clues to assist with the backup issue please.

I want to upgrade from OS X 10.5.8 to Snow Leopard, then to Lion. When trying to back-up to a WD 1 TB external drive, the back-up process stalls at about 4 GB download. I would be very grateful for any clues to assist with the back-up process please.

That disk is very probably formated as FAT 32, the normal format for PCs. FAT 32 cannot support files larger than 4 GB; also, it corrupts some types of Mac OS files. If you want to securely backup your mac files, you must have a disk prepared for Mac OS.
Therefore, open Disk Utility, choose again 1 partition for your external disk and, under Options, GUID partition map (it may also be a boot disk for an intel mac), then format it as Mac OS Extended Journaled. Y0u are done.
If you want to exchange data with PCs, you should chose Master Boot REcord and NTFS format, for this you must use a additional utility like NTFS-3G. But note that mac files may also get corrupt in NTFS format.

Can Time Machine back up Fibre Channel connected drives?

Hello All,
     I am running Mac OS X 10.7.5. I am trying to back up drives connected to my tower via fibre channel using Time Machine and am getting the exclusion error that others have been reporting on Lion and Mountain Lion. In Time Machine > Options, I select a fibre channel connected drive to include it in the next backup. It disappears from the exclusion list as it should AND the properly "estimated size of full backup" is created. I select "Save." However, when I select "Options" again, the fibre channel connected drive has been RE-EXCLUDED. This happens with all of the drives connected via fibre channel.
     All of the drives have unique UUIDs and display Connection Bus: Fibre Channel Interface and Connection Type: External in Disk Utility. All of the drives are formatted as "Mac OS Extended." The drives are all inside of a Facilis Terrablock 24D and are RAID5.
      Before updating to Lion, we were able to back up these drives via Time Machine successfully for many months.
      Does anyone know if it is possible to back up drives connected via fibre channel with Time Machine on Lion and Mountain Lion? To be perfectly clear, I am trying to simply back up these fibre channel connected drives to a Firewire connected external HD via Time Machine...NOT use a fibre channel connected drive as my Time Machine backup disk.
       Your time and help is greatly appreciated!
Best,
LALB

Others have reported the same issue. Apparently it doesn't work.

Easiest solution for adding Fibre Channel Tape Backup to Xserve RAID setup

I'm interested in getting a Fibre Channel LTO-3 tape library such as the Exabyte 221L for my Xserve RAID/Xserve G5 setup. I realize there are SCSI LTO-3 solutions that are less expensive than FC, but I'd prefer not to have to add a SCSI card to my Xserve, mainly because I'd have to pull the video card and run the server headless (horrors!). However, I understand that I must get a Fibre Channel switch to allow the LTO library, the Xserve, and the Xserve RAID to communicate... you can't just daisy chain FC devices. The least expensive switch I can find is the new QLogtic SANbox 3050 for about $1800. For my situation, in which these will be the only FC devices I'll probably ever use, does this switch sound appropriate? The specs on the switch can be found here: http://www.qlogic.com/products/sanbox/sanbox_3050.asp.
Xserve RAID & Xserve G5 Mac OS X (10.4.5)

For simple environments like yours, it's trivially easy. You must plug everything in, and you're good to go... as simple as a network switch.
Now if you start adding multiple hosts (i.e. plug in a second Xserve or G5), you need to be careful, because in the default configuration everything is "wide open," so every device can see every storage LUN. This would actually be problematic, as systems could (actually, would) corrupt each other's storage. When you get to this point, you'll need to look at either zoning the ports on the switch, or using LUN masking on the Xserve RAID to allocate specific pieces of storage to specific hosts.
But for now, just get the switch, plug everything in, and you're off.
BTW if you mark "answered" for me, versus for the thread overall, it helps me out a bit in this forum ranking system, whatever it is.
Cheers.

Xserve, G5, Xserve RAID and fibre channel question

hi guys,
I have a Xserve conneted to Xserve RAID right now. Is it possible connect G5 to Xserve RAID thru fibre channel ? thanks

>is it possible to do ?
Not in the way you describe, no.
The XServe RAID's 14 drives are connected to two separate controllers. Drives 1-7 are connected to the upper controller. Drives 8-14 are connected to the lower one.
If your XServe is currently connected to the upper controller it's talking to drives 1-7.
If you connect the G5 to the lower controller it will be talking to drives 8-14.
Therefore you cannot use this setup to backup the data via the G5.
At the very, very least you need a fiber channel switch so that both the XServe and the G5 can connect to the same controller. However, as Tod has already mentioned, it is tricky to get multiple machines talking to the same RAID drives while avoiding corruption (there's no concept of arbitration, or ways of preventing one machine from editing a file that's in use by the other machine).
You have two basic options - one is to implement XSAN, which will handle the arbitration issue, the other is to configure the G5 to mount the RAID read-only. This should prevent the G5 from modifying a file while it's in use by the XServe (which may be sufficient for backup purposes), but you may still run into issues when the G5 tries to backup a file that's actively in use by the XServe.
In all honesty, a better solution would either to get a SCSI card that works with the XServe and do it all there, or use file sharing between the XServe and the G5 and backup over the network - it'll be slower, but a lot more reliable.

Clarification on how to use Xserve Raid and Fibre Channel without xsan.

First let me apologize for not responding earlier to your response, I tend to get busy and then forget to check back here.
Tod, the answer to your question is No, only one computer is accessing the xserve raid files at any one time and that is via Fibre Channel. However I do have the xserve raids set up as share points via ethernet.
Maybe I should turn that off and only access the files with the one computer that can connect via fibre channel.
I never thought of that. I will try that while I await for your answer, thanks again.
Todd Buhmiller
I have the following setup:
Xserve: 2x2Ghz Dual Core Intel Xeon, 5Gb of Ram, Running 10.5.8 Leopard Server
Xserve Raid with firmware version 1.5.1/1.51c on both controllers, and
Qlogic Sanbox 5600
Apple Fibre Channel Cards in Xserve, and Mac Pro Tower; Apple 2 Port 4Gbs Fibre Channel Card
Mac Pro Tower-Quad Core Intel Xeon 2.8Ghz, 16Gb of Ram, Running Snow Leopard 10.6.4
Here is the problem.
The directory for the xserve raids keep getting corrup, and I use disc warrior to rebuild them. Is there a way to keep the directories from getting corrupt? I am a few pieces of equipment before I can build an Xsan as that is the ultimate goal, but until then, I just need to be able to have the raids funciton as storage without having to rebuild the directories all of the time.
Anybody have any suggestions?
Thanks
Todd Buhmiller
Widescreen Media
Calgary, Alberta Canada
Tod Kuykendall
Posts: 1,237
From: San Diego
Registered: Oct 11, 2000
Re: Xserve Raid Mounts, Corrupt Directory tired of rebuilding directory
Posted: Jun 27, 2010 1:25 PM in response to: Todd Buhmiller
Are multiple computers accessing the same data on the RAID at the same time?
If so then NO. This is the source of your data corruption and I'm surprised if you were able to get all your data back every time if this is how you've been running your system. Each fibre channel assumes it has full and sole control of every volume it has mounted, no data arbitration is practiced and data corruption will occur if this assumption is wrong.
The only way this set-up will work is to use partitions or LUN masks so the volumes are accessed by one computer at any time. As long as one computer relinquishes control before another mounts it you will dodge arbitration issues but this is a dangerous game. If you screw up and mount an already mounted volume - and there is no easy way to tell if a volume is mounted - corruption will then occur. Sharing data simultaneous at fibre speeds is what XSAN does and to do this you need it.
HTH,
=Tod
Intel Xserve, G5 XServes, XRAID, Promise

+The xserve raids will mount automatically to any computer that I connect the qlogic fc switch to+
This is source of the corruption to your data. Any computer that attaches to a drive/partition via fibre channel assumes that it alone is in control of the drive and data corruption is inevitable.
+Is that the issue, should I disconnect the xserve from the fc switch and leave it connected via ethernet?+
Short answer: YES. The ethernet connections are fine because the server is controlling the file arbitration through the sharing protocol. Fibre channel connections assumes complete control over the partition and no arbitration of the file access is performed. It's like two people independently driving trying to drive the same car to different locations.
Depending on your set-up it is possible for the two machines to see and use different parts of the Xserve RAID storage but they cannot access the same areas without SAN doing the arbitration.
Hope that's clear,
=Tod

RMAN with Netbackup - ALL Tape backups expired

Hello,
I am currently experiencing some issues between Netbackup and RMAN.
A bounce of the Netbackup server seems to have marked all our tape backups managed by RMAN as 'Expired'. This means RMAN can no longer locate any of the tape backups.
We have checked the rentention periods and the physical backups on the tapes and they seem to be ok. Therefore it would suggest a link problem between Netbackup and RMAN. We have tried bouncing the Netbackup servers again but this has not resolved the situation.
However sometimes the tape backups become available and but then become expired again.
I have also tried backing up to Tape since the bounce and the backup is successful, however 24 hours later it becomes expired despite Retention policy set to 4 weeks on both RMAN and Netbackup.
Please help!
Thank you

Hey;
I tried the delete obsolete; command initially. Admittedly, I didn't report that - so here's the output:
RMAN> delete obsolete;
using target database control file instead of recovery catalog
RMAN retention policy will be applied to the command
RMAN retention policy is set to recovery window of 3 days
allocated channel: ORA_DISK_1
channel ORA_DISK_1: sid=137 devtype=DISK
allocated channel: ORA_DISK_2
channel ORA_DISK_2: sid=132 devtype=DISK
Deleting the following obsolete backups and copies:
Type Key Completion Time Filename/Handle
Backup Set 6 10-SEP-09
Backup Piece 6 10-SEP-09 c-1369681761-20090910-00
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of delete command at 07/24/2010 21:36:58
RMAN-06091: no channel allocated for maintenance (of an appropriate type)
As for the changing default device type: someone had mentioned that the default device type was disk so I changed it to tape, tried the command, failed, then switched it back.
Just to reiterate: I don't have a tape drive on this box; never did. I'm not sure how two tape backups could have gotten there in the first place.
The oracle version is:
Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - 64bit Production
With the Partitioning, OLAP and Data Mining options
and the OS version is:
$ uname -r; cat /etc/redhat-release
2.6.18-128.el5xen
Red Hat Enterprise Linux Server release 5.3 (Tikanga)
If push comes to shove, I can leave those two backups. There were a rather large number of them that I did successfully make go away. It just seems odd that I can't make these guys go away.
Thanks for the help; I appreciate it.
Doug

Xserve/XRAID to HP Tape with Retrospect. Problems?

We have a quad core xeon xserve connected via fibre to an XRAID with ~2TB of data. If the data is lost we're in trouble. We have a used HP StorageWorks 1/8 Ultrium 960 Tape Autoloader we can use for tape backups. I found that EMC Retrospect can work with this particular model, but we'll need to purchase a SCSI card to connect it to the Xserve.
My question is; What problems, if any, should I anticipate having with this configuration? I've heard SCSI can be a nightmare, but is it ok under Leopard Server? Is Retrospect the best/easiest option here? Anything else you can add with regards to this would be very helpful. Thanks

Ultra 320 SCSI? I tried to install the card, but I misunderstood the manual which led me to believe I had a PCI-X and PCI-E riser card in the server. Apparently since we got it from Apple with the fibre card it came with two PCI-E risers. So, we now have to order the PCI-X riser in order to install it.
The card came with no real documentation beyond the pages from the server manual showing how to install the card and that to configure a SCSI card you would need to use Disk Utility. At any rate the tape autoloader has a terminator on its second port. Would that normally be enough or would I need another one on the expansion card's second connector too?
And a terminator isn't brand specific is it? I would imagine we have some laying around from HP hardware we use. Would that be usable?
Thanks

Xserve RAID Fibre Channel Lights pegged until disk activity

We are having some issues with an Xserve RAID connected to a G5 Xserve running 10.4.8 Server. The Fibre Channel lights on the front are constantly maxed out, but whenever the disk are active the lights die down only to peg again when activity stops. Needless to say, we are having SEVERE access speed issues. Any ideas?

This can happen in a protected RAID set (i.e. RAID 5) if a single raid member drive pops offline and back on again, since the RAID set has to re-build itself from the parity data in that case.
It also happens after creating a RAID 5 volume, as the creation usually takes a few days (assuming you used the 'RAID Now' option), and so you'll see the blue lights.
It also makes sense that the blue light 'spiking' would drop when you do a read/write, as the initialization process has to pause to handle the disk activity.
Are either of these the case?
We are having some issues with an Xserve RAID
connected to a G5 Xserve running 10.4.8 Server. The
Fibre Channel lights on the front are constantly
maxed out, but whenever the disk are active the
lights die down only to peg again when activity
stops. Needless to say, we are having SEVERE access
speed issues. Any ideas?
MacBook Pro Mac OS X (10.4.8)

Performance issue in tape backup

Friends,
Our tape backup is implemented with help of Veritas Netbackup,for one of our servers,tape backup is taking too much long time. I generated AWR report during the ongoing backup and noticed the below stats:-
Top 5 Timed Events
Event
Waits
Time(s)
Avg Wait(ms)
% Total Call Time
Wait Class
Backup: sbtwrite2
36,186
300,497
8,100
1,442.5
Administrative
CPU time
17,920
85.4
Backup: sbtbackup
38
1,892
48,000
9.0
Administrative
log file parallel write
112,632
1,378
12
6.6
System I/O
control file parallel write
57,983
1,075
19
5.1
System I/O
What will be the next troubleshooting steps ?
Any guidance will be highly appreciated.
Thanks in advance.

The tape drive is likely the constraint. It cannot write fast enough to handle 4 channels streaming datafiles to it.
I would guess that the datafiles are also going over the network to the media manager. You need to check the network throughput for that link as well.
Hemant K Chitale

Problem with fibre channel card and raid system after upgrading to 10.4.7

Hello, I have a problem with our Xserve G4 DP. After the installation of 10.4.7 everything works fine exept our fibre channel card. This card is connected to a Hitachi Raid System which worked fine over a year. But after the update the volume was no longer mounted. I even reinstalled the server to 10.4.4 and no sign of our raid volume. Using the apple fibre channel utiltiy 2.0, I see that there is a connection - but still the raid volume is not mounted. When I restart the server the lights on the back of the card are shortly flickering and sending some packets to the raid system but that´s all. Has anybody an idea?
Greetings from Heidelberg
Klaus
Xserve G4 DP 1 GHz Mac OS X (10.4.6) Mac OSX Server 10.4.7

You could re-update, by downloading the 10.4.7 Combination Updater from Apple.

Xserve Raid will not mount - fibre channel link down

I came back from being out of town and my Xserve is running, but will not mount on the G5 tower via fibre channel. It was powered off and up again and still nothing. Apple tech says to reset raid controllers, however, I have sensitive data on the drives and lots of video/FCP projects on there. It has just run out of warranty and now asked to pay $199 per incident to get any help. I will gladly take any help.
Read the boards, ran Disk Utility and still did not show up.
Also, told the fibre cables tend to "burn out" but have yet to spend money if this is not the problem.
Thanks.

Resetting the controllers is non-destructive. However, if you power-cycled the array, that should take care of it.
told the fibre cables tend to "burn out"
I've never experienced (or heard of this) myself, and have never had a problem on any of my arrays, some of which have been running for years with zero problems. Of course, environment (especially heat) could play a part, but it seems a little bit of a stretch to me.
Does RAID Admin have anything to say about the RAID?

4 xserves, 1 xserve raid, 4 partitions, fibre channel - possible?

Is it possible to connect 4 xserves to 1 xraid that has 2 partitions for each side of the controller to all connect together using a fibre channel switch?
Thanks!
Tony

Hello Paul,
LUN slices are all equal sized - ie you slice it into two, three or four slices but they will all be equally sized. You can't slice off 1/3 of a RAID side and the leave the other 2/3s alone. (I guess you could slice it into three and then RAID0 the other two slices back together but that seems too tricky to be a good idea.) The advantage is that each LUN has a logical "name" and you can hardwire your fibre card to see only that "name" and there is zero chance of conflicting control of a slice.
If I understand you correctly you want to RAID 50 the entire RAID into a single big volume and the partition it up into smaller logical partitions. You will need to do the RAID0ing of the two sides before anything else because this will destroy the data that already exists on the RAID 5s. As long as you keep this in mind it is doable and you can then use fstab editing to control which Xserves load which volumes on start-up. Note that volume partitioning is not a set-up for a hostile environments because people will be able to do bad things if they are determined to do so - eg Using Disk Utility to force mount a volume they otherwise can't see and then play with it.
My question is "Do you really need a single large volume?" Alot of people do RAID 50 the unit into a single image but unless you need more than 3TB volume sizes for some reason or for HD video capture I personally don't see the attraction. You take RAID 5 which is pretty close to bullet proof but then you build it into a RAID 0. The data on a RAID 0 is susceptible to loss from breaking the RAID while two separate RAID5s (especially with hot spare drives) is about as close as you can get to rock solid in computing. If you're going to then divide it up and redistribute it anyway I'm not sure if the merging of the sides is necessary.
Only you can determine what you need but if you're not going to have a single >3TB volume I would go RAID5s on each side (7drives or 6+1spare ideally) and then either slice or partition up the volumes for the respective computers.
My $.02,
=Tod
G5/2.0x2, Dual XServes x2, XRAID, beige G3 501Mhz

Fibre Channel Link light off on Xserve

I noticed a few days ago that the lower Fibre Channel link light and Activity lights are no longer working on the lower unit on my Xserve. So far I tried unplugging and replugging the Fibre cable but that didn't help.
THere are no warning lights lit on the Fibre card on the back of my G5, and from flipping through the Xserve manual, it looks like all those lights are correct also.
We recently had the network rewired and since then I haven't been able to access the RAID Admin to see what's up. I may have to reset the controller though.
If you have any advice on where to start troubleshooting, or settings to check once I get the RAID Admin working I'd appreciate the advice, thanks!
PS: I have installed the Xserve Driver Update 1.0

Everything seems to be working normally. I'm mastering off the few projects I just finished, and I'm able to play back to streams of Uncompressed 10 bit SD so far so good. Although the one working channel seems to be working a bit harder than usual
The Xserve RAID is hooked directly to my powermac G5 via Apple's Fibre Channel Card. I talked to my boss who has been working wit the network people and said the IP address should be the same, and it should be picking up DHCP. From your post it sounded like there may be a setting in RAID Admin for this? I'll check after I finish Edit to Tape
Think of my calmness this way: If the front of your car is hanging off the edge of a cliff, the last thing you'd want to do is make any sudden moves!

Xserver Fibre Channel with tape backup issues?

Similar Messages

Maybe you are looking for