• Priorities

Analyses

Two issues:

1. Providing better feedback to user regarding the status of their jobs so they don’t sit and wait for it to finish.
2. How to speed up jobs.

Queuing –

Better interface to make the job transparent to the user so they don’t sit and wait.

Giving performance study a higher priority so the jobs aren’t blocked by long analyses. Tom – has some reservations about the usefulness of performance study.

• Threshhold times – We need to avoid recalculating the threshold times. Should we change upload to include it? We can make it asynchronous. Threshhold times or split script error. There used to be a hash at the beginning to of the threshold file to indicate that we would have to rerun all past data. When and how would we rerun threshold times? We need to write a design spec. on how we imagine solving this. We used to cache them. All the .thresh files were stored with the data files. We need to do timing tests. Tom could run a series of workflows against canonical files to see how much time they are actually taking.
• Improving code as well. Testing should look at places in the code that are taking more time than they should.

Blessing

Issue 1: Rolling out new version

Currently, www13 has blessing 2 – automatic blessing with benchmark files (this is different from how to handle legacy data). Tom wants a date for rolling out blessing 2 onto production – automatic blessing and hiding data that is not blessable. Bob said fellows are looking; we need to gather their opinions.

He suggested Nov. 1 as drop-dead date for e-Lab fellows to have completed their feedback on blessing. Then we can study the comments and make any changes needed and put them into production.

Issue 2: retroactive blessing.

There was a long discussion of Tom’s method that does not require a benchmark file. Liz suggested that we might want to have a way to distinguish between data blessed using benchmark files vs. that blessed using Tom’s method for legacy data. (Silver star vs. gold star – for blessing ). Bob reminded us that the site just gets more and more complicated.

Notifications

We need to look at what Edit has added on www13.

Problem with Computing Division Resetting power settings and missing data while machine goes to sleep. Increase the frequency of the STDs. Service desk ticket to make an exception.

CMS e-Lab

Prioritizing with Tom McCauley what needs to be done in the CMS e-Lab.

Implementation of the poster

Publishing of Posters

Now we have the publishing of posters. We need a simple rubric whether a poster should be published or not. Talk with the fellows that used the e-Lab to get their idea what they would consider to be publishable or not publishable. Ken will talk with the LHC e-Lab fellows. Distill the essence of a good poster. Bob says only 28 teachers used it.

Wiki – dependencies

Liz talked about how we need to rid of them. (e.g., pages that are transcluded from mediawiki). Ken still has content in wikis, but this is different from the content Liz mentioned. She wrote a document she gave Edit about all the places we need to fix.

Other Topics

Property Records Change:

We have to reassign ownership of the cluster machines at Argonne because Edit is not a lab employee now. Bob? Liz?

Ken – Tokyo workshop. Nov. 9^th and 10^th.

We may need to think about the design of the workshop. Blessing and flux. (Tom L. wanted the performance plot as part of the Blessing plots. ) Do we need to design something in the procedure to avoid the crunch when analysis jobs are not getting completed. Bob focuses more on performance and blessing plots. Skip over plateauing. Ken wants Bob and Tom to look at what he proposed. Cheryl Brooks is going to Japan, too. You need to have something in your hip pocket when the server his hung up. How can we minimize the issue. (e.g. four people at a computer)?. Have one computer to demo on for everyone to see. When they do their projects, they naturally submit them asynchronously. Do not do more than one days worth of data. No lifetime and shower, usually.

-- Main.LizQuigg - 2013-10-07