TSM 6.2.3 + Vstorage + ESX 4.0: Strange restore

[mitch08]

Hello all,

I have a problem to restore a full VM, it's slow! (128mb by 128mb every two minute)
I have this message in the tsm log (Cannot use mode hotadd).
....Is it a problem?

[2011-04-06 10:55:05.515 02592 verbose 'transport'] CheckHotAddLicense: has beencalled
[2011-04-06 10:55:05.562 02592 verbose 'transport'] CheckHotAddLicense: We are connected to a server and not a Virtual Center
[2011-04-06 10:55:05.562 02592 verbose 'transport'] The host ha-host is not licensed for HotAdd
[2011-04-06 10:55:05.578 02592 error 'transport'] This VM [class Vim::VirtualMachineStub:00000000024591B0] is not licensed for HotAdd
[2011-04-06 10:55:05.578 02592 verbose 'transport'] Mounting VM vim.VirtualMachine:544 using transport hotadd failed : HotAdd is not licensed.
[2011-04-06 10:55:05.578 02592 verbose 'transport'] Cannot use mode hotadd to access [Storage1] Vcenter_1/Vcenter_1.vmdk: Cannot mount using this method.
[2011-04-06 10:55:05.578 02592 info 'transport'] Creating mount directory C:\DOCUME~1\ADMINI~1\LOCALS~1\Temp\vmware-Administrator\564d0e06-ab46-0fe7-946e-e7d96fe135a3-544\nbdssl.
[2011-04-06 10:55:05.593 02592 info 'transport'] Starting to write unmount info.

..sorry for my bad english

Thank you.

 

[mitch08]

I progress, it's strange....
fisrt test:
- backup in STG Pool TAPE > Ok
- restore > very slow (128mo by 128mo 500ko/s)

second test:
- backup in STG Pool DISK > OK
- restore > very fast (40 mo/s)

What's the problem ?

 

[mitch08]

I check the contain of the tape:
The store size is 128 Mo ....??

tsm: TSM>q cont A00102 f=d

Node Name: FULL_VM_DAILY
Type: Bkup
Filespace Name: \VMFULL-par-mv2prod
Hexadecimal Filespace Name:
FSID: 1
Client's Name for File: \par-mv2prod\SNAPSHOT_015000000_20110404190106\par-mv2prod.ovf
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 6,224
Segment Number: 1/1
Cached Copy?: No

Node Name: FULL_VM_DAILY
Type: Bkup
Filespace Name: \VMFULL-par-mv2prod
Hexadecimal Filespace Name:
FSID: 1
Client's Name for File: \par-mv2prod\SNAPSHOT_015000000_20110404190106\Hard Disk 1\JOB015000001\MBLK0000.DAT
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 128.02 M
Segment Number: 1/1
Cached Copy?: No

Node Name: FULL_VM_DAILY
Type: Bkup
Filespace Name: \VMFULL-par-mv2prod
Hexadecimal Filespace Name:
FSID: 1
Client's Name for File: \par-mv2prod\SNAPSHOT_015000000_20110404190106\Hard Disk 1\JOB015000001\MBLK0001.DAT
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 128.02 M
Segment Number: 1/1
Cached Copy?: No

Node Name: FULL_VM_DAILY
Type: Bkup
Filespace Name: \VMFULL-par-mv2prod
Hexadecimal Filespace Name:
FSID: 1
Client's Name for File: \par-mv2prod\SNAPSHOT_015000000_20110404190106\Hard Disk 1\JOB015000001\MBLK0002.DAT
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 128.02 M
Segment Number: 1/1
Cached Copy?: No

Why 128 Mo?

My option file:

NODENAME FULL_VM_DAILY
PASSWORDACCESS GENERATE
TCPSERVERADDRESS *.*.*.*
SCHEDMODE POLLING
SCHEDLOGNAME "C:\progra~1\Tivoli\Tsm\baclient\dsmsched_VMDAILY.log"
ERRORLOGNAME "C:\progra~1\Tivoli\Tsm\baclient\dsmerror_VMDAILY.log"
SCHEDLOGRETENTION 7
ERRORLOGRETENTION 7
COMPRESSION NO
DEDUPLICATION NO
VMMC "MC_VMWARE_DAILY_ONTAPE"
VMBACKUPTYPE FULL
VMFULLTYPE VSTOR
VMCHOST par-vcenter01
VMCUSER administrateur
VMCPW ****

 

[mitch08]

I call the support IBM and open a case....(go to the lab)

I downgrade the client 6.2.3 to 6.2.2 and there is not problem anymore!!

Node Name: NAT-PROXYTSM
Type: Bkup
Filespace Name: \VMFULL-par-test
Hexadecimal Filespace Name:
FSID: 21
Client's Name for File: \par-test\SNAPSHOT_001000000_20110411092440\par-test.ovf
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 5,357
Segment Number: 1/1
Cached Copy?: No

Node Name: NAT-PROXYTSM
Type: Bkup
Filespace Name: \VMFULL-par-test
Hexadecimal Filespace Name:
FSID: 21
Client's Name for File: \par-test\SNAPSHOT_001000000_20110411092440\Hard Disk 1\JOB001000001\MBLK0000.DAT
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 20,482.50 M
Segment Number: 1/1
Cached Copy?: No

Node Name: NAT-PROXYTSM
Type: Bkup
Filespace Name: \VMFULL-par-test
Hexadecimal Filespace Name:
FSID: 21
Client's Name for File: \par-test\SNAPSHOT_001000000_20110411092440\Hard Disk 1\JOB001000001\MBLK0000.NCTL
Hexadecimal Client's Name for File:
Aggregated?: No
Stored Size: 382
Segment Number: 1/1
Cached Copy?: No

 

[smetter]

Thanks for sharing this information Mitch.

I had the exact same problem and also solved it by downgrading the client back to 6.2.2.

It's a pain in the ass though,
because I was going to test the new TDP for VMware,
which requires BA client 6.2.3 to perform block level incremental snapshots, individual file restores from snapshots, VM disk drive mounts from snapshots, etc.

I guess we will have to wait a bit until baclient 6.2.4 is there (or some intermediary bugfix)

 

[mitch08]

Hi Smetter,

The next TSM version to perform block level incremental snapshots, individual file restores from snapshots, VM disk drive mounts from snapshots, etc. is....Tivoli Storage Manager for Virtual Environments. It's a new TDP with a new licence! (No trial version)

It's a pain in the ass though

....I badly understood...The bug continues with the TDP + baclient 6.2.3 !??

 

[smetter]

I would think so since the TDP uses the functionalities provided by the BA client API.

I have discontinued implementing the TSM for VM TDP because of this, so I didn't test this.

 

[dynemix]

Hi..

Looks like I'm experiencing the same problem just like you both did..

Is there any update from IBM on this ? because I'm planning to use TDP for VE too...

 

[smetter]

Not that I know of.

I'm waiting for BA client 6.2.4 / 6.3

 

[shribar]

I had the same problem with client 6.2.3.1 and solved it with additional mgmtclass for control files that goes to disk(file) device class.

https://www-304.ibmw.com/support/docview.wss?uid=swg27021081

http://www.ibm.com/developerworks/w...ts-Configurethebackuparchiveclientoptionsfile

 

[smetter]

Thanks for sharing shribar.
I've also implemented the extra management class with success,
works like a charm!

 

[francs]

Thanks for the suggestions.
This is a major problem.
We didn't have capacity in our library so we implemented DP for VE and have been using it for 2 months now, doing incremental backups.
All that money spent for nothing.

I've tried the management class solution and it helped, but only a little.
My vmware image backups go straight to tape and the control info goes to disk now.
I can see a lot of simultaneous open/close input volume for disk and tape, which confirms that it works.
However, I'm only getting a restore speed of around 24Gb/hr. (At least it's better than the 3Gb/hr I got before).
Back to 6.2.2 I guess

 

[francs]

mithch08
Have you had any feedback from the call you logged with IBM regarding this problem?

 

[mitch08]

Hi,
Yes, this option was missing in dsm.opt

VMCTLMC "DIRMC"

 

[walkman]

i use 6.2.2 BA client backup to VTL storage pool. backup was ok but problem with restore.. it's take to long to complete restore process.