djciaro
Expert
Expert

Backup of RAW disk via HP Data protector

Hi Guys,

I have the following issue when trying to use HP Data Protector to run a full backup of a windows server running in an ESX 3.0.2 Cluster.

Background info:

2 HP Proliant Blade BL20p with 2 FC cards QLA231x/2340

All virtual machines are running on SAN

1 of the Windows servers has its own SAN which is presented via ESX as a RAW disk.

Before the migration to Virtual the backup of the data on SAN lasted about 5 hours.

600GB full copy every night

After migrating to VI and updating the Data protector agent, presenting the Tape device to the VM via SCSI and editing the backup job to reflect the new settings everything seemed fine.

However the problem starts after the backup starts, at first the copying of data runs at about 50,000 kbps but after few mins it slows down to about 25,000 and continues to slow down until it runs at about 8,00o kbps.

This means that a backup that used to run in 5 hours takes about 15 hours to complete and as a result the rest of the jobs remain in the queue and the backup window is used up by one backup job.

For company policies an incremental or differential backup is not permitted, it has to be full everyday.

I have been checking the settings on the ESX and changing some of the settings to see if it is an issue with the vmHBAs or switches.

at the moment the SAN for VM storage has 4 paths (2HBAs and 2 Switches) 1 path preferred vmHBA 0:0:1:1 & 3 standby

The server that has the SAN presented is using a different path vmHBA 1:1:2:5

The disks on the server have High shares (2000) all of the other servers have normal (1000) shares.

I am unsure if the problem is because of the settings on the ESX or if data protector needs something changed as the connection between the backup device goes via an ESX server and not directly as with a physical server.

If anyone has any ideas or is running backups of SAN devices that are presented to a VM as a RAW disk I would greatly appreciate any tips or advice.

HP DP version is 5.5

ESX 3.0.2 VC 2.0.2

I have attached a screen shot of the disk performance on the VM when the backup starts, it illustrates the way the reads drop suddenly after 10 mins.

thanks in advance

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!
0 Kudos
3 Replies
djciaro
Expert
Expert

Hi Guys,

I have been working on lot on this issue to try and find a work around, I have applied some patches to the ESX servers to resolve the issue of snapshots with RDM disks and also upgraded to VC 2.0.2 U1.

Now I can connect a RAW disk in physical mode and a snapshot works fine. ESX 1002088 & ESX 1002090

so I figured if I change the SAN that is presented to the windows server as RAW disk but in Physical mode instead of virtual mode then maybe I can get the backup to work via the FC and run at normal speeds.

Now the situation is that the backup starts very fast (reading 50,000KBPS) and after 10 minutes it starts to drop until it has copied 40GB and then it fails with errors such as "cannot write to device" and Scsi a errors, which gets me thinking that sometime during the backup the connection between the tape device and SAN is lost. resulting in time outs and errors.

I am using a HP MSL 3060 connected to SAN and the ESX servers are HP blades BL20P with Qlogic FC adapters (QLA231/2340)

I have presented the Tape device to the Windows VM and everything is step up correctly in Data protector.

Does anyone know of any changes that I might need to apply to the Qlogic cards or SAN settings on the ESX server, is there a possibility that there is a buffer size that needs to be increased, also does anyone know if VMware supports this kind of a backup.

Windows 2003 SP2 Standard ed VM running on an ESX 3.0.2 cluster with VC 2.0.2.

SAN is presented as a RAW disk in physical mode and works fine at domain level (access, speed, etc)

The backup is with DP 5.5 and is a task that needs to backup about 600GB

Tape devices are presented to VM as SCSI devices

starts fine and suddenly the speed drops until it fails.

Thanks again

ref attached error from DP

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!
0 Kudos
BUGCHK
Commander
Commander

VMware does not support Fibre Channel-attached tape drives.

Did you disable Windows' own Removable Storage Manager?

It will periodically send TUR (Test Unit Ready) commands.

I think the Data Protector documentation describes it.

djciaro
Expert
Expert

Hi,

Thanks for replies: I am having trouble understanding if VMware does not support Fibre-Channel attached tapes, how is it possible that I can get as far as 50GB of data copied over to the library via a FC connection?

When the SAN configuration documentation says: *Dedicated Adapter for Tape Drives . For best results, use a dedicated SCSI adapter for any tape drives that you are connecting to an ESX Server system. FC connected tape drives are not supported.*

Does that mean it does not work (full stop) or that it may work it is just that they have not fully tested it yet and therefore will not commit to supporting it.

The HP documentation suggests:

Windows 2000/2003 tape drive polling

The Windows Removable Storage Manager service (RSM) polls tape drives on a frequent basis (every three seconds in Microsoft Windows 2000, and every second in Windows Server 2003). This polling, which involves sending a Test Unit Ready (TUR) command to each tape or library device with a loaded driver, is enabled by way of the AUTORUN feature of Windows Plug and Play. Windows built-in backup software (NTBACKUP) relies on the device polling to detect media changes in the tape drive.

Note

In SAN configurations, this polling can have a significant negative impact on tape drive performance. For SAN configurations, HP strongly recommends disabling RSM polling.

For additional information, refer to the Microsoft Knowledge Base article number 842411.

There is a broken link in the MS KB to a HP site

Solution for HP devices

HP provides a workaround for the TUR request issue that you experience in Windows Server 2003. For more information about this workaround, visit the following HP support Web site:

so I cannot see what that might suggest.

As I have no access to the customers servers until tomorrow I cannot verify if the removable storage service is disabled, however I do not that by default it is disabled and I cannot think of any reason why it would have been enabled on this server.

Do you think it would be possible to use the proxy server as a backup device destination.

i.e. the proxy server has a FC connected to the switch and could be used in the HP DP backup job as the destination.

At the moment I am using the tape devices that are presented to the VM via ESX and as SCSI devices not FC.

To be honest all of this is causing a big issue as the whole project is being seen as a failure just because of 1 backup job when in fact migrating all of the servers to VI3 have been a great success.

It would be really great if i could close the chapter on this issue.

I really find it hard to believe that nobody else has come accross this issue.

A windows server that has SAN presented in physical mode, you want to make a file level backup via an MSI 6030 connected and with no other space available in SAN to replicate the data. how can you back it up? ?:|

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!
0 Kudos