drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From 风间苍月 <zbd...@163.com>
Subject Could a individual sys.profiles table be useful for drill?
Date Thu, 22 Dec 2016 13:34:47 GMT

I am a ordinary Drill user managing a cluster of Drill. In our environment the cluster serves
a lot of queries every day, so since the Drill web provide 100 latest query profiles (and
without any failure filtering there), we are not able to analyze earlier queries which are
eventually failed from execution. 

For solving this problem, firstly I deployed a ELK cluster for managing all drill query logs,
but this solution is finally too heavy. After that, I made a sys.profile implementation to
Drill source, it is something like this:

This change is already enough for my original requirement. Could this thing be helpful in
most cases? If yes, I have already filed a PR referring to the JIRA issue:


  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message