p10101 not on psummary

[harlam357]

As it says... p10101 is no longer on the psummary page. I'm am still receiving these WUs on my GPU2 clients, so they must still be active.

I have reports from HFM users who have installed a new instance of HFM and cannot get a reading for this Project. I now see why, it's not on the psummary.

[smoking2000]

You should cache the projects listed on the psummary pages locally, so they're always available in your application.

Your application cache can be synced with the Jmol fah-projects, FCI project-summary and fahinfo.org export data

[harlam357]

I do. But I don't currently deploy the application with a pre-built "cache file" - due to these recent issues I've been considering it.

It's not automated (yet) but I occasionally upload a new ProjectInfo.tab file here (just updated it): http://hfm-net.googlecode.com/svn/trunk ... ctInfo.tab

Jmol looks pretty complete, but the core id strings seem to be truncated to a "code" tag == more conversion for me. The fahinfo.org data doesn't list these p101xx projects either - and it's just in a totally different format. I like how your FCI data looks, but it looks like the kfactor data is incomplete. I agree, it would be nice if we all had a consistent data source.

I'll have to think about this some more. I may just stop pulling directly from the psummary and just auto update my file and pull from there. We'll see.

Caching data or not... the bottom line is - why aren't Projects currently in circulation listed on the psummary pages???

[bruce]

... the bottom line is - why aren't Projects currently in circulation listed on the psummary pages???

I can think of several possibilities. (I don't have access to any of the actual code, so these are just speculation from the external symptoms.)

1) I suspect that the Psummary script polls the servers periodically but does not cache data. This means that if a server is off-line at the moment it is polled, the data probably disappears until the next time a poll finds that project back on-line. Meanwhile the server, itself, can come back on-line and distribute work.

2) A new version of the psummary scxript has been deployed. It may have a bug.

3) New server code is being deployed. This might have caused unforeseen issues in the polling process . . . especially since the psummary script has to work simultaneously with servers that have either the old or the new server code.

4) The integration of the various servers into a single system is really very complex and it runs 24x7x52. Making changes to a real-time system is fraught with pitfalls and there's really no way to test that integration in a laboratory environment.