[
https://issues.apache.org/jira/browse/SOLR-7605?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14570032#comment-14570032
]
Hoss Man commented on SOLR-7605:
--------------------------------
In all of these failures, PivotFacetField.createFromListOfNamedLists, which is
used to construct PivotFacetField trees from shard responses, is getting called
with a list of pivot values for "field=dense_pivot_ti" which includes multiple
instances of "field=dense_pivot_ti,value=0,count=0"
*Even though "0" is _NOT_ a valid value for the dense_pivot_ti field in any
document* ... this (multi-valued) field is only ever populated with random
values between 20 and 50!
NOTE: These are the per-shard requests with mincount=0, so _if_ "0" was a valid
value in this field, a single instance of count=0 wouldn't be that weird.
Things i've made notes on while testing out these seeeds:
* all failing seeds so far include 3 instances of value=0,count=0 at end of list
* in some seeds stats are hanging off of these pivot values, in some seeds
there are no stats
* in some seeds the dense_pivot_ti field is a sub pivot, in others the top
field, sometimes only field, in the facet.pivot
* various posting formats are used for dense_pivot_ti:
PostingsFormat(name=Direct), PostingsFormat(name=LuceneFixedGap),
BlockTreeOrds(blocksize=128)
here's the "top level" queries in involved in each seed:
{noformat}
[junit4] 2> 42725 INFO
(TEST-TestCloudPivotFacet.test-seed#[22F85D14F0CCB183]) [ ]
o.a.s.c.TestCloudPivotFacet Doing full run:
{main(facet=true&facet.pivot=%7B%21stats%3Dst3%7Dpivot_d&facet.pivot=dense_pivot_ti%2Cdense_pivot_i%2Cpivot_b1&facet.limit=13&facet.offset=6&facet.pivot.mincount=188),extra(rows=0&q=*%3A*&fq=id%3A%5B*+TO+894%5D&stats=true&stats.field=%7B%21key%3Dsk1+tag%3Dst1%2Cst2%7Dpivot_tl&stats.field=%7B%21key%3Dsk2+tag%3Dst2%2Cst3%7Dpivot_i1&stats.field=%7B%21key%3Dsk3+tag%3Dst3%2Cst4%7Dpivot_z_s&_test_min=188)}
[junit4] 2> 37444 INFO
(TEST-TestCloudPivotFacet.test-seed#[7A1923556F2286C2]) [ ]
o.a.s.c.TestCloudPivotFacet Doing full run:
{main(facet=true&facet.pivot=%7B%21stats%3Dst3%7Ddense_pivot_ti%2Cpivot_b1&facet.pivot=%7B%21stats%3Dst1%7Dpivot_tl1%2Cpivot_f%2Cpivot_x_s1&facet.limit=9&facet.offset=7&facet.pivot.mincount=1678),extra(rows=0&q=*%3A*&_test_min=1678)}
[junit4] 2> 31392 INFO
(TEST-TestCloudPivotFacet.test-seed#[3D83C25D5DB9AE70]) [ ]
o.a.s.c.TestCloudPivotFacet Doing full run:
{main(facet=true&facet.pivot=%7B%21stats%3Dst1%7Dpivot_td%2Cpivot_z_s1%2Cpivot_d1&facet.pivot=%7B%21stats%3Dst1%7Ddense_pivot_ti&facet.limit=13&facet.offset=5&facet.pivot.mincount=423),extra(rows=0&q=*%3A*&_test_min=423)}
[junit4] 2> 34191 INFO
(TEST-TestCloudPivotFacet.test-seed#[F0A169D7E9F4C1C2]) [ ]
o.a.s.c.TestCloudPivotFacet Doing full run:
{main(facet=true&facet.pivot=%7B%21stats%3Dst3%7Ddense_pivot_y_s%2Cdense_pivot_ti&facet.pivot=%7B%21stats%3Dst2%7Dpivot_dt1%2Cdense_pivot_y_s%2Cpivot_x_s&facet.limit=17&facet.offset=6&facet.pivot.mincount=801&facet.overrequest.count=5),extra(rows=0&q=id%3A%5B*+TO+553%5D&fq=id%3A%5B*+TO+1127%5D&stats=true&stats.field=%7B%21key%3Dsk1+tag%3Dst1%2Cst2%7Dpivot_y_s1&stats.field=%7B%21key%3Dsk2+tag%3Dst2%2Cst3%7Dpivot_dt1&stats.field=%7B%21key%3Dsk3+tag%3Dst3%2Cst4%7Dpivot_tdt1&_test_min=801)}
{noformat}
---
On the assumption that maybe there was a problem resulting in "0" being
returned by the (Uninverted) DocValues when a document has no value in a field,
and/or when the precisionStep injected some synthetic terms, I made some quick
attempt to reproduce the problem manually by hand crafting some similar shard
queries against a single node solr instance with a small amount of data -- but
i've still had no luck.
So now i'm going to capture the indexes from the shards of a failed seed,
manually run servers using that data & configs, and try to reproduce the
duplicate "0" values by querying them directly.
> TestCloudPivotFacet failures: Must not add duplicate PivotFacetValue with
> redundent inner value
> -----------------------------------------------------------------------------------------------
>
> Key: SOLR-7605
> URL: https://issues.apache.org/jira/browse/SOLR-7605
> Project: Solr
> Issue Type: Bug
> Reporter: Hoss Man
> Assignee: Hoss Man
>
> There have been two recent jenkins failures of TestCloudPivotFacet on both 5x
> and trunk with the same underlying cause...
> {noformat}
> <p>Problem accessing /collection1/select. Reason:
> <pre> java.lang.AssertionError: Must not add duplicate PivotFacetValue
> with redundent inner value</pre></p>
> {noformat}
> ..digging through mail logs, it looks like there have been a handful of these
> errors on different branches and os, with and w/o nightly, since April 1st of
> this year.
> The two recent seeds i tried (on trunk and 5x) reproduce - details to follow.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
