You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Ishan Chattopadhyaya (JIRA)" <ji...@apache.org> on 2017/03/07 13:28:38 UTC

[jira] [Comment Edited] (SOLR-10178) TestInPlaceUpdatesDistrib unable to use NoMergePolicy[Factory] on branch_6x

    [ https://issues.apache.org/jira/browse/SOLR-10178?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15899427#comment-15899427 ] 

Ishan Chattopadhyaya edited comment on SOLR-10178 at 3/7/17 1:27 PM:
---------------------------------------------------------------------

Indeed, this was same as in master as well.

I realized that the following would always use a MergePolicyFactory:

{code}
    systemSetPropertySolrTestsMergePolicyFactory(NoMergePolicyFactory.class.getName());
    System.setProperty(SYSTEM_PROPERTY_SOLR_TESTS_USEMERGEPOLICYFACTORY, "true");
    System.setProperty(SYSTEM_PROPERTY_SOLR_TESTS_USEMERGEPOLICY, "false");
{code}

It seems that doing this is the only way to always use the NoMergePolicyFactory, since we cannot use NoMergePolicy. This hack is at least better than "skip the troublesome code" using assumeFalse. Still, I'm unhappy about having to hack this way; but unfortunately, I don't have time to dive deeper right now.


was (Author: ichattopadhyaya):
Indeed, this was same as in master as well.

I realized that the following would always use a MergePolicyFactory:

{code}
    systemSetPropertySolrTestsMergePolicyFactory(NoMergePolicyFactory.class.getName());
    System.setProperty(SYSTEM_PROPERTY_SOLR_TESTS_USEMERGEPOLICYFACTORY, "true");
    System.setProperty(SYSTEM_PROPERTY_SOLR_TESTS_USEMERGEPOLICY, "false");
{code}

It seems that doing this is the only way to always use the NoMergePolicyFactory, since we cannot use NoMergePolicy.

> TestInPlaceUpdatesDistrib unable to use NoMergePolicy[Factory] on branch_6x
> ---------------------------------------------------------------------------
>
>                 Key: SOLR-10178
>                 URL: https://issues.apache.org/jira/browse/SOLR-10178
>             Project: Solr
>          Issue Type: Test
>      Security Level: Public(Default Security Level. Issues are Public) 
>            Reporter: Ishan Chattopadhyaya
>
> TestInPlaceUpdatesDistrib depends on consistent segments to track docIds to assert in-place updates.
> Towards that, NoMergePolicy is best suited and working fine on master (by defining it using NoMergePolicyFactory).
> This doesn't work with just the MergePolicy (systemSetPropertySolrTestsMergePolicy()), since NoMergePolicy is a singleton and doesn't have a constructor. Setting only a NoMergePolicyFactory (systemSetPropertySolrTestsMergePolicyFactory()) seems to take no effect and falls back on RandomMergePolicyFactory. As a result, this test cannot use the NoMergePolicy[Factory].
> Here's the corresponding test failure: https://jenkins.thetaphi.de/job/Lucene-Solr-master-Linux/19012/
> -Seems to me that SOLR-8668 needs to be backported to branch_6x for this test to work, or some other stopgap hack needs to be put in place to make it work before SOLR-8668 is backported. [~cpoerschke], WDYT?-
> Edit: This is also valid for master. ^



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org