Finished Phase One

I've finished the first phase of performance testing. Everything is cpu bound in the Zope client. The Zeo server never breaks a sweat and i/o in negligible. The next step is to find out what part of the code is hogging the cpu. I need some Zope permission changes before I can check that.

More Performance Testing

I ran the copy-paste and delete tests today and got similar results to yesterday. Zope is badly compute bound.

Testing Zope Performance

I spent the day testing zope performance, doing parameterized runs of adding methods to a folder. It turns out that the zope client is strongly compute bound. Each process does take up a good chunk of memory (100 Mb), but only two or three processes are enough to fully use the cpu.

More tests

I finished testing the Event Manager. I also did some testing of Zope performance. I didn't see the horrible lags when saving the file that I have previously, either on the test server or the production server. I'm wondering if the lag is a function of the moon phase.

Testing and Scoping

I did some more testing of the Event Manager and then stopped, because it seems the email notification is not configured properly on the test server. All tasks that send email hang and the mail is never received. Since I'll have to retest anyway when the problem is fixed, it seemed better to stop testing for now.

So I turned to working on the scoping document. I finished the second version and sent it to Marty. That work is complete unless he asks for changes.

Testing

I spent the morning working on the scoping and the afternoon testing Event Manager. I got through most of the test, but there are still four pages to go.

Stuff

I started the day testing zope performance. I switched over to testing the event manager for Greg and when I got stopped on that, I switched over to testing my modifications to the ldap report writer. The Perl script now works, but the TeX I am telling it to produce is still wonky.

Scoping LDAP Project

I wrote a first draft of the ldap user authentication project scoping and sent it around for comment.

End of the Week

While I was able to bring up the test instance of zope yesterday, it crashed after running for a short while. I guessed that the problem was that the version of the Data.fs I copied used different products than those used in the zope client. I copied the zope database from baragon and all seemed to be well. I started up the test instance of squid and it seemed to work fine. However, when I tried to connect to www.stsci.edu from ubik, it went to the real zope rather than my test instance of zope. That's the next hurdle I have to jump.

In the mean time, I talked with Ron and Dan about authenticating computer users with ldap. It looks like keepeing separate directory servers running for each platform and replicating between them is the way to go.

I fixed the error in my revised ldap reporting script that will be able to generate the staff list, but I have yet to test it. I made a small change in the comment describing the divisional administrator in the ldap manager to excise the hated word "secretary."

Setting Things Up

I spent most of the day trying to get the test zope and zeo server up and running. It seems that we succeeded but there may still be problems. I'll know more tomorrow. I also spent some time scoping out the proposed ldap authentication project and working on the revised reporting script to produce the staffling list from ldap. The latter has a small problem computing the fields to sort on.