Present: Liz, Tibi, Eric, Nick and Bob
Load Balancing:
Bob wants a deadline to get the load-balancing cluster on the air and "offering". Needed before the first week of July and many e-Lab workshops.
LIz indicated that www10 can be added to the load-balancing pool.
In order to get load-balancing to work properly we have to solve the login issue.
Tibi will read Liz’s document on logging in where she suggests a solution that will allow bookmarked student home and teacher pages to work because the login form on the home page could take them to a login on the apache server that will process the e-lab name and redirect them to a login for that e-Lab on one of the load-balanced machines. Her document is accessible on the EntryPoints page. Tibi needs to experiment and see if it is possible.
Bob brought up the problem of people getting errors "page not found" because of bad bookmarks. Eric suggested implementing a 404 page – Error message that points them to the correct place. We can look at a referrer as well.
When Tibi has the load-balanced pool ready to go, we need to hammer on it.
Load-balancing may help searching somewhat but we only have one database.
Analysis and Rollout
Nick and Tom added a lot of changes that need to be rolled out. These include lots of error checking in geometry entry and better input including which channels are active. These need rolling out. Because Tibi is going to Texas, he will try to roll out to www14 today for Nick to test and then rollout to 11 as soon as possible if everything is o.k.
Eric's Report:
Eric took us for a tour of some of the Wiki items he has added. In particular he showed us how News is implemented on his ligo site and pointed us to his notes on how to set up an RSS feed. RSS feeds don't work in IE6. We could use an include file on the home page that would include either the dynamic content from the news or a static html page that could be made each time we add a news item. We did not have time to look at his news management system.
He showed us some notes on his analysis system as it is implemented in PHP. Liz pointed out that the analysis will need to be converted to the approach used for e-Labs for it to be compatible grid submission and the VDS. That is one reason Mike hoped for him to get his code running on one of the i2u2 machines, independent of the e-Lab.
We had to stop the meeting after 45 minutes due to other obligations. Eric will continue his tour next week.
Not discussed in Meeting, but added as news.
Understanding Differences between Rollout to www13 and www11
We discovered that rolling out to www13 is different from www11 because when a rollout is done, it actually involves a cvs update of the files that are served. When we work on www13, we tend to copy over the files in the served area and this can really mess up CVS. For now, we will not rollout to www13 directly as we do with www11 and the other test machines. We don't actually have a proper rollout for this so it is hard to test saving plots.
www13 is no longer very sluggish, especially on login (DNS?)
-- Main.LizQuigg - 07 Jun 2007