slackhappy commented on a change in pull request #1436:
URL: https://github.com/apache/lucene-solr/pull/1436#discussion_r473305818
##########
File path: solr/solr-ref-guide/src/common-query-parameters.adoc
##########
@@ -206,7 +206,7 @@ The default value of this parameter is blank, which causes
no extra "explain inf
== timeAllowed Parameter
-This parameter specifies the amount of time, in milliseconds, allowed for a
search to complete. If this time expires before the search is complete, any
partial results will be returned, but values such as `numFound`,
<<faceting.adoc#faceting,facet>> counts, and result
<<the-stats-component.adoc#the-stats-component,stats>> may not be accurate for
the entire result set. In case of expiration, if `omitHeader` isn't set to
`true` the response header contains a special flag called `partialResults`.
+This parameter specifies the amount of time, in milliseconds, allowed for a
search to complete. If this time expires before the search is complete, any
partial results will be returned, but values such as `numFound`,
<<pagination-of-results.adoc#using-cursors,`nextCursorMark`>>,
<<faceting.adoc#faceting,facet>> counts, and result
<<the-stats-component.adoc#the-stats-component,stats>> may not be accurate for
the entire result set. In case of expiration, if `omitHeader` isn't set to
`true` the response header contains a special flag called `partialResults`.
Review comment:
I was just adding to the existing list there. I'll pull it out into its
own sentence.
In my opinion, `timeAllowed+cursorMark+omitHeader` should be allowed, and
SOLR shouldn't be second-guessing the caller's request:
1. The combination `shards.tolerant+cursorMark` has the same issue, and
`shards.tolerant+cursorMark+omitHeader` is allowed.
2. There may be a reasonable scenario that simply doesn't call for it, even
if it is advisable to keep the header. Let's say you are implementing a
product catalog, and:
- For whatever reason, you don't need the header, you rely on the http
status code
- You want to use cursorMarks to allow people to scroll through your
fine list of shoes or whatever
- You want to time-bound that resultset and protect your infrastructure
from long-running queries
- It just simply isn't mission critical for that every user to see every
shoe. If some rare expensive query or latency-producing event occurs, and
truncates your shoe list, it is okay (you probably wouldn't bother retrying if
you knew partialResults were true anyway).
My feeling is also a bit of a reaction to being annoyed about SolrCloud
preventing us from creating collections without replicas, because we had
legitimate reasons for standing up the replicas ourselves (thankfully, there is
a workaround).
Then again, I say this as someone who doesn't have to answer Solr customer
questions, so my words are just words.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
