PSSP Disaster Recovery

[1500wing]

Hello,

I have an SP Frame running PSSP v3.4 with 6 Winterhawk Nodes. I want to use an F50 as a Disaster Recovery machine and back the nodes up over ethernet, rather than the SP switch. AND, have the DR F50 seen\"managed" by the current CWS. Management doesn't want us to put an unknown load on the switch, even thought the backup will occur late evening, early morning.

What is the best way to do my DR with this F50 over ethernet for the 6 production PSSP nodes? The CWS is also an F50, if that matters. My migration\installation documentation doesn't cover this, or I m ay have missed it.

Thanks - Chris

Thanks
-Chris

 

[screwloose]

I am unclear what the purpose of the second F50 is being used for. Is it a repository for a mksysb from the client nodes?

You can add a second ethernet card to the second F50 so you can run some PSSP commands to it, but it cannot be SP-attached like an S80.

 

[1500wing]

The second F50 was acquired to be the backup node for the prod nodes, via the ethernet connection, using ADSM, then spitting it out to tape. <sigh> The powers that be, wish to use the second F50 as a backup and DR for the nodes, using ADSM. Sounds like a nightmare...at least in a SP scenario.

Personally, I have suggested that we backup the nodes via mksysb images to the CWS, since it appears we have enough disk space, then just rotate the backup images every couple of days, so all nodes have two backups on the CWS. OR, just add the second F50 as a secondary CWS, since we do not have a spare CWS right now.

Since it sounds like the second F50 cannot be attached like a SP node, it might be a good boat anchor? Any way around this? Thanks!!

 

[screwloose]

I am currently a PSSP admin for a 6 node complex, however my previous job was as a PSSP admin for a 70+ node SP complex with a 1600 Cluster (with a p670 and H80's).

Our DR procedure was to run a mksysb on the nodes and then ftp the images to the control workstation (/spdata/sys1/images) where they could be used for recovery, if needed.

During DR testing we would install the mksysb’s to the nodes and that would include the script to use to rebuild the VG’s, LV’s, FS’s, etc. The script was executed and then after this step was done, then a dump was made from TSM (ADSM previously) to restore the data.

It worked very well and our DR tests were successful.

As far as a second CWS, don’t add the second F50 in an HACWS environment. That is more hassle than it is worth. If something goes wrong with the CWS, just restore the mksysb and data, etc. As long as the nodes are healthy you can get by without your CWS until it is working again.

I am not as sharp as I used to be either, so if you could spell out for me what exactly you or they want to do from this comment, “The second F50 was acquired to be the backup node for the prod nodes, via the ethernet connection, using ADSM, then spitting it out to tape.”, then maybe I can come up with something for you to do, but really the way we did it worked very well as I mentioned.

 

[1500wing]

Hello, there is no HA environment within the PSSP complex, so at least I don't have to worry about that yet. For some odd reason, the Admin(s) prior to my arrival here, wanted to purchase the second F50, have it "managed" by the current CWS, yet leave it as a "standalone" server, in other words do not add it to the current production frame. The plan is to use ADSM to back it up over the private ethernet network every night.

Personally, it doesn't make a lot of sense to me, since in my past history, I did the same thing as you, back up the mksysb images to the CWS and archived them as well.

My suggestion is, either use this F50 as a semi-expensive space-heater, or add it as the secondary CWS for the PSSP frames. Thanks!

 

[Breslau]

You can also automount the images directory when you write your mksysb's.

putting ML levels in the filenames helps too. i've twice had to retreat on ML's after applying them broke something else (like LTO).

Cheers

 

[1500wing]

Thanks for the input!

-Chris