chudi
Active Newcomer
As verified (advices):
- dsmgui (graphical interface of hsm) always search the opt file (dsm.opt) in the install directory (default: C:\Program Files\Tivoli\TSM\hsmclient). Doesn't work to invoke dsmgui indicating the path of opt file (CMD> C:\Program Files\Tivoli\TSM\hsmclient\dsmgui -optfile=G:\dsm.opt)
- Some programs (TSM BA) one of each change the access date when make backup. You have to add "preservelastaccessdate yes " to the dsm.opt of BA, this option avoids this change by BA.
- For delete the filespace created by HSM use nametype option (DELETE FILESPACE HNODE_NAME FILE_SPACE NAMEType= FSID or UNICODE).
Conclusion:
Unable to change the configuration directory of HSM opt file (dsm.opt) is not possible to implement the classic clustering like for BA.
Proposal:
I have a cluster, 3 physical machines and 5 virtual. Is necessary to install HSM client on all physical machines (3). Then I register each nodes: clu_1, clu_2 and clu_3. In the HSMs dsm.opt of 3 nodes I put the next option "asnodename cluster_hsm." Cluster_hsm have to be registered and grant proxy_node permissions to clu_X nodes. Always HSM will manages all the disks as local disk, also all the information of the 3 nodes will be the property ofcluster_tsm. Finally, HSMs jobs should be generated in the 3 teams.
Doutbs: Any other option .. ?
- dsmgui (graphical interface of hsm) always search the opt file (dsm.opt) in the install directory (default: C:\Program Files\Tivoli\TSM\hsmclient). Doesn't work to invoke dsmgui indicating the path of opt file (CMD> C:\Program Files\Tivoli\TSM\hsmclient\dsmgui -optfile=G:\dsm.opt)
- Some programs (TSM BA) one of each change the access date when make backup. You have to add "preservelastaccessdate yes " to the dsm.opt of BA, this option avoids this change by BA.
- For delete the filespace created by HSM use nametype option (DELETE FILESPACE HNODE_NAME FILE_SPACE NAMEType= FSID or UNICODE).
Conclusion:
Unable to change the configuration directory of HSM opt file (dsm.opt) is not possible to implement the classic clustering like for BA.
Proposal:
I have a cluster, 3 physical machines and 5 virtual. Is necessary to install HSM client on all physical machines (3). Then I register each nodes: clu_1, clu_2 and clu_3. In the HSMs dsm.opt of 3 nodes I put the next option "asnodename cluster_hsm." Cluster_hsm have to be registered and grant proxy_node permissions to clu_X nodes. Always HSM will manages all the disks as local disk, also all the information of the 3 nodes will be the property ofcluster_tsm. Finally, HSMs jobs should be generated in the 3 teams.
Doutbs: Any other option .. ?