I get information per client session and per file read ... so I can select individual jobs based on "global" properties of the file access.

Looking at plots of duration the file was open and the fraction of file was read, I see that if I select files that read close to 100% of the file and are open for 1000 - 3000 s I get about 6% of the total number of jobs -- these might be the successful ones. This is the bump in the top-right of the lego plot below.

Now, this isn't really the access pattern CMS sees for analysis jobs on AOD / MiniAOD data, it's more likely to read some fraction of data (as not all branches are accessed), can be anything from 10 to 60%. On NANO/MICRO AOD one would expect the file to be read in full. Do you know what data-tier we have here and if it is expected that the job would read every branch in the file?

open_duration
frac_read
title

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.

[ { "@context": "http://schema.org", "@type": "EmailMessage", "potentialAction": { "@type": "ViewAction", "target": "https://github.com/xrootd/xrootd/issues/1259#issuecomment-673744245", "url": "https://github.com/xrootd/xrootd/issues/1259#issuecomment-673744245", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { "@type": "Organization", "name": "GitHub", "url": "https://github.com" } } ]

Use REPLY-ALL to reply to list

To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1