SPARC meeting minutes

SPARC Meeting Minutes Archive

January 14, 2009

Tying up loose ends so far:

• Remaining putbacks
   - Thanks Joe for putting back
   - Thanks Sue for taking 5787 (bootroot rename), and Karen and Joe for helping out.
• Status on nightly builds
   - Clay will be getting these going.
   - We need a SPARC image built ASAP for testing.

Unit-Testing with the "real bits":

• People are to start testing, installing ldoms virtual machines and a common current AI SPARC image to be made available by Clay and Mary. osol-inst will be used as the image server. This is unit testing by developers, before handing off the image to the AI test team next Monday.
• Test machines have been moved to the AI local network.
• How to setup an LDOMS machine
   - Clay has posted instructions.
   - MAC addresses from OK prompt.
   - IP address can be assigned thru AI. Mary or Clay will make available the internal IP addresses to use for the ldoms.
• Logistics of getting the "real bits":
   - Clay to get with Mary to post the AI image on osol-inst, the AI install server.
   - Mary needs instructions on how to update osol-inst. Sundar will help. Sundar to send out instr to everyone.
   - Mary has concerns about OpenSolaris and Solaris together on the same subnet.
   - She will install on sun4v without ldoms, before releasing image to everyone (internally).

Roundtable

• Joe: another known ICT bug: 6080 ICT set_Solaris_partition_active needs to handle OBP device names
   - Joe to work with Tycho.
   - Will target completion for this Friday.
   - If it goes in this Friday, we can give to AI test people from the start, after doing a sanity test.

December 17, 2008

 Project statuses:

 - AI

  Sundar has been testing booting from net, running autoinstall, using images created by DC. Can download microroot.

 Wasn't setting up service names, so couldn't completely run with UI engine.

 Not completely automated yet. Will try running end to end by setting up services today manually.

 Needs wanboot binary as part of iso. Need to point to it in config file; it needs to be downloaded first, followed by the bootroot.

 William is working on AI engine and orchestrator changes. From his status given Fri, can run everything up to the point where the slim_install package is needed.

 David C is working on getting the SPARC repo to be closer to the X86 repo than it was (e.g. will have slim_install cluster soon)

 Jan's issues around DCFS will be handed by DC changes.

 Jan sees two issues with respect to populating device tree:

 1) lock file in /etc

 2) live-devices-local script in slim_source calls devfsadm for liveCD images but not for AI images. For SPARC (not for X86), this script needs to call devfsadm for AI images too.

 Joe reported that Tycho doesn't think there is much more to do for getting ICT library in shape for SPARC than what Jan already found, and should be a simple matter to code. (He is still checking though.) The main issue is setting up an environment where they can manually issue ICT requests to insure that they work properly. Tycho and Joe will get together to set up this environment.

 Sue's installadm tools are progressing and will be done in time.

 Testing should be completed the second workweek in January.

• • - SPARC team 

 Already reported above.

• • - DC 

 Karen: SPARC support requires 3 outstanding changes:

 1) Have DC to call installboot.

 2) wanboot.

 3) fiocompress.

 Jean putback changes to build AI image for SPARC.

 It is noted that the current "ISO" for AI won't boot. This can be rectified in the future if it needs to be. Currently AI uses the ISO as a big container, and extracts from there what it needs.

• • - Snap 

 Evan reports that it's all done! Testing complete!

• • Machine update 

 Probably time to scrap the idea of using osol-bldsp as a build machine; the larger disks we had didn't work. Since Clay reports that there is a new sun4v machine in Santa Clara, we'll try that one. It has ample memory, disk space. Compute power should be good, assuming there is parallelism during builds. The only question is how much slower it will be because it is not in MPK. Clay to set this up and evaluate after the break, since buxton.central is still available.

 LDOMs is being worked now. Clay is most of the way there; is having trouble creating clients. Greg offered to help him out.

• • Overall SPARC project status 

 SNAP status is: done for January.

 DC status is: mostly done except for cleanup and tweeking... and there is still some tweeking to do.

 AI is coming together and will be done and ready for formal testing by mid January. The missing pieces have been identified and are being worked.

 Testing will be ramping up as development ramps down. LDOMs (virtual machines) running on sun4v machines should be ready by the time we return from break, in time for meeting heavier needs.

 Build machine will come on line soon, but has lower priority for Clay than LDOMs as buxton.central is still working.

• • Roundtable 

 Thanks to everyone for your dedication to this effort!

 There will not be a meeting next week or the week after. Next meeting will be on Jan 7. Happy holidays!

 Thanks,

 Jack