[Isocops] Moving to SwRI

[fair-play at comcast.net]

Roland, 


We didn't get as far as starting a list - you've made a great start! 


Some thoughts on your list: 


1. Although it's up to Nathan, I think we should get the SwRI machines to the point where they can do full data reduction and also get the web site up and running there. It would be a good test to see if we can use those machines as the primary. My motivation: major upgrades in the AC and perhaps UPS are supposed to start in the machine room at UNH in the late summer to fall timeframe. I expect that there will be times when the machines may have to be shut down gracefully (and not so gracefully). If there are any problems/accidents during construction we may find it easier to run at SwRI for a bit. 


2. Your questions about the hardware configuration at SwRI are all good. By the way, I was able to mount one of the RAID partitions so it looks like the RAID is at least present. 


3. Thanks for pointing me to those documents. I think it would be very helpful for RCC to have an updated version of this document. The time we spend in updating it will probably be well spent in fewer hassles as we move machines to UNH. 


4. I think your idea of moving off Geoff's account and to some general account for processing is the way to go. 


5. Same with your suggestions for the processing tree and release. 


6. We will have to get SwRI to allow access from UNH so the exercise of allowing access from somewhere other than ena will be good practice. 


7. I did a scan through the code and found these cases where host names are hard-coded: 


misc_update.sh 
nightly.sh 
ssr_blk_update.sh 
sftp_driver.sh 
grab-data.sh (testing) 
misc_cleanup.sh 


Host names are also mentioned in id-base and id-leops. In addition, there are a few places where mail is sent to either Geoff or ibexops. I'm certain Geoff will want to continue to receive mail but we may want to replace the hardcoded mail addresses with an environment variable or two. 


I'm sure I'll think of more... 




Ken 








----- Original Message ----- 
From: "Roland Vanderspek" <roland at space.mit.edu> 
To: isocops at lists.sr.unh.edu 
Sent: Thursday, June 17, 2010 2:12:55 PM GMT -05:00 US/Canada Eastern 
Subject: [Isocops] Moving to SwRI 




All, 


There has been discussion of preparing to move Ops to SwRI as part of the transition to UNH. You probably discussed this at the ISOC telecon; I apologize for having to ask questions you've probably already answered. 


I think it's important to plan (at least), start (probably), and maybe even finish (hopefully) this process before Geoff goes pumpkin in two weeks: there are too many details which we'll need his brain for to try to do it without him. I have a couple of questions regarding the plan, and then I'd like to start making a checklist of the things we need to do to make the transition. It would be good to converge on the plan before the meeting next week. 


If such a checklist already exists, stop reading here and please tell me where to find it! 


Questions, mostly for Nathan: 




    • What functionality are we trying to install at SwRI? I can see ops-only, or that and any of a number of levels of data reduction, or full data reduction. 
    • Will a web presence at SwRI be required? sweb.swri.edu ( https://129.162.237.196/ibex/ ) is visible, but there's no there there. Will we be expecting this level of functionality during this (supposedly brief) transition period? 
    • Is there a schedule for when this should happen? 
    • Do the IT folks at SwRI know this is going to happen? 
    • What is the hardware configuration at SwRI these days? Was its infrastructure upgraded to the mirror system at BU when BU was changed? A quick "mount" on swena reveals no RAID mounted. Swena is at 129.162.237.195, sweb at 129.162.237.196, and there's a swoc at 128.197.73.227. 



Ken, please refer to the files ISOC_HW_config.txt and ISOC_HW_svn.pdf in /usr/local/ibex/sw/isoc/src/doc/Admin for an overview of the hardware at the time of its writing. Changes to the configuration were made a year or so back, and these changes probably did not make it into this document, but it's a very good start. We should probably look at putting it into texi format and including it in the ISOC manual. 


Things that have to be done (partial list, in no particular order for now): 




    • Decide which account is doing the routine ops: currently it's gbc, will it stay that way? I think it should transition to ibexops or another account not bound to one of us. 
    • Decide which tree is used for routine ops: currently it's gbc, we'd discussed making it the most recent release. Doing all analyses with the latest release makes the most sense to me, going forward. 
    • Find and replace with environment variables all hard-wired " ena.bu.edu " et al. references: review files /usr/local/IBEX/ops/config/isoc_hw.config and ISOC.raid. 
    • Define a release 4.0 of the ISOC tree and install at SwRI. If we decide that running from the latest release is the correct paradigm, do it on ena first. 
    • Allow outside access to swena assuming ena.bu.edu is down (this may take some talking to the IT folks at SwRI). 
    • Get the id-leops and id-base areas on swena (and sweb) up to date. Is this a brute-force copy of the data on ena, or do we let swena do a grind of all the raw directories and then compare results? I suggest it should be the latter. 
    • Get the SwRI hardware to the point where it's "identical enough" to the BU setup to get the job done. 



I'm sure Geoff can add megalines to this list without batting an eye. Anyone who has a thought should add it. Once we have a more complete list, we need to organize it in time order and assign some actions. 


Roland 


Please restrict discussions on this email list to non-ITAR sensitive topics. ______________________________________________ Isocops mailing list Isocops at lists.sr.unh.edu http://lists.sr.unh.edu/mailman/listinfo/isocops
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sr.unh.edu/mailman/private/isocops/attachments/20100618/c1404987/attachment-0001.html