[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[ https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16767015#comment-16767015 ] Nitin Gupta edited comment on OAK-8035 at 2/13/19 10:09 AM: [~tmueller] - yup - I don't have commit rights . This is actually my first contribution here. So I will be glad if you or [~teofili] can commit for me . Thanks a lot for the help . was (Author: nitigup): [~tmueller] - yup - I don't have commit rights . This is actually my first contribution here. So - yeah you or [~teofili] can commit for me . Thanks a lot for the help . > Debug logging when two or more indices have same or very close cost amounts > doesn't work in case both indices belong to the same type of Query Index > > > Key: OAK-8035 > URL: https://issues.apache.org/jira/browse/OAK-8035 > Project: Jackrabbit Oak > Issue Type: Bug > Components: query >Reporter: Nitin Gupta >Assignee: Nitin Gupta >Priority: Major > Attachments: OAK-8035.patch, OAK-8035_2.patch > > > Steps to reproduce - > # You would need two index definitions of same query index type , let's say > fullText index , having similar cost evaluations. > # Now while the index plan is evaluated you will notice that there should > have been a debug log saying that selected index has similar cost - this > serves as an important warning to either change the index definitions or the > query > ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)] > > However If we have a look at this code block here - > [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041] > , there is no comparison for near best costs for index plans having same > query index type . > > cc : [~teofili] -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[
https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16769438#comment-16769438
]
Julian Reschke edited comment on OAK-8035 at 2/15/19 4:08 PM:
--
With this change, the build fails for me:
{noformat}
[WARNING] Corrupted STDOUT by directly writing to native stream in forked JVM
1. See FAQ web page and the dump file
/mnt/c/projects/apache/oak/trunk/oak-core/target/surefire-reports/2019-02-15T15-44-24_303-jvmRun1.dumpstream
[ERROR] Tests run: 42, Failures: 0, Errors: 18, Skipped: 0, Time elapsed:
80.928 s <<< FAILURE! - in
org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest
[ERROR]
testImportWithRestrictionNodeInfo(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 9.922 s <<< ERROR!
java.lang.OutOfMemoryError: Java heap space
[ERROR]
testImportWithRestrictionNodeInfo(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 9.922 s <<< ERROR!
java.lang.NullPointerException
[ERROR]
testEndChildInfoIncompleteAce(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 5.238 s <<< ERROR!
java.lang.OutOfMemoryError: Java heap space
{noformat}
Backing out the change is fixing he build for me.
was (Author: reschke):
With this change, the build fails for me:
{noformat}
[WARNING] Corrupted STDOUT by directly writing to native stream in forked JVM
1. See FAQ web page and the dump file
/mnt/c/projects/apache/oak/trunk/oak-core/target/surefire-reports/2019-02-15T15-44-24_303-jvmRun1.dumpstream
[ERROR] Tests run: 42, Failures: 0, Errors: 18, Skipped: 0, Time elapsed:
80.928 s <<< FAILURE! - in
org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest
[ERROR]
testImportWithRestrictionNodeInfo(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 9.922 s <<< ERROR!
java.lang.OutOfMemoryError: Java heap space
[ERROR]
testImportWithRestrictionNodeInfo(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 9.922 s <<< ERROR!
java.lang.NullPointerException
[ERROR]
testEndChildInfoIncompleteAce(org.apache.jackrabbit.oak.security.authorization.accesscontrol.AccessControlImporterIgnoreTest)
Time elapsed: 5.238 s <<< ERROR!
java.lang.OutOfMemoryError: Java heap space
{noformat}
> Debug logging when two or more indices have same or very close cost amounts
> doesn't work in case both indices belong to the same type of Query Index
>
>
> Key: OAK-8035
> URL: https://issues.apache.org/jira/browse/OAK-8035
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: query
>Reporter: Nitin Gupta
>Assignee: Nitin Gupta
>Priority: Major
> Fix For: 1.11.0
>
> Attachments: OAK-8035.patch, OAK-8035_2.patch
>
>
> Steps to reproduce -
> # You would need two index definitions of same query index type , let's say
> fullText index , having similar cost evaluations.
> # Now while the index plan is evaluated you will notice that there should
> have been a debug log saying that selected index has similar cost - this
> serves as an important warning to either change the index definitions or the
> query
> ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)]
>
> However If we have a look at this code block here -
> [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041]
> , there is no comparison for near best costs for index plans having same
> query index type .
>
> cc : [~teofili]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[ https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16769620#comment-16769620 ] Julian Reschke edited comment on OAK-8035 at 2/15/19 7:27 PM: -- trunk: [r1853659|http://svn.apache.org/r1853659] [r1853626|http://svn.apache.org/r1853626] was (Author: reschke): trunk: [r1853626|http://svn.apache.org/r1853626] > Debug logging when two or more indices have same or very close cost amounts > doesn't work in case both indices belong to the same type of Query Index > > > Key: OAK-8035 > URL: https://issues.apache.org/jira/browse/OAK-8035 > Project: Jackrabbit Oak > Issue Type: Bug > Components: query >Reporter: Nitin Gupta >Assignee: Julian Reschke >Priority: Major > Fix For: 1.12, 1.11.0 > > Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch > > > Steps to reproduce - > # You would need two index definitions of same query index type , let's say > fullText index , having similar cost evaluations. > # Now while the index plan is evaluated you will notice that there should > have been a debug log saying that selected index has similar cost - this > serves as an important warning to either change the index definitions or the > query > ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)] > > However If we have a look at this code block here - > [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041] > , there is no comparison for near best costs for index plans having same > query index type . > > cc : [~teofili] -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[ https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771731#comment-16771731 ] Julian Reschke edited comment on OAK-8035 at 2/19/19 8:58 AM: -- [~nitigup]: FWIW, I find it unfortunate that we introduce a new test dependency here ([http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-core/pom.xml?r1=1850878&r2=1853626&pathrev=1853626&diff_format=h).] Is this really needed? And if it is, shouldn't we use the latest version? (see [https://mvnrepository.com/artifact/org.assertj/assertj-core]) was (Author: reschke): FWIW, I find it unfortunate that we introduce a new test dependency here ([http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-core/pom.xml?r1=1850878&r2=1853626&pathrev=1853626&diff_format=h).] Is this really needed? And if it is, shouldn't we use the latest version? (see https://mvnrepository.com/artifact/org.assertj/assertj-core) > Debug logging when two or more indices have same or very close cost amounts > doesn't work in case both indices belong to the same type of Query Index > > > Key: OAK-8035 > URL: https://issues.apache.org/jira/browse/OAK-8035 > Project: Jackrabbit Oak > Issue Type: Bug > Components: query >Reporter: Nitin Gupta >Assignee: Julian Reschke >Priority: Major > Fix For: 1.12, 1.11.0 > > Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch > > > Steps to reproduce - > # You would need two index definitions of same query index type , let's say > fullText index , having similar cost evaluations. > # Now while the index plan is evaluated you will notice that there should > have been a debug log saying that selected index has similar cost - this > serves as an important warning to either change the index definitions or the > query > ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)] > > However If we have a look at this code block here - > [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041] > , there is no comparison for near best costs for index plans having same > query index type . > > cc : [~teofili] -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[
https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771825#comment-16771825
]
Julian Reschke edited comment on OAK-8035 at 2/19/19 11:17 AM:
---
{quote}[~reschke] - Just find it a bit more fluent with assertions - makes the
test code a bit more clean to look at and I used the version that I had used
before to make sure it works as I expect . But I can remove it - I also need to
refactor the test to use the LogCustomizer as you suggested - so will remove
this dependency also along with that.
{quote}
Done that already, and I don't thik the test is harder to read. See
[r1853872|http://svn.apache.org/r1853872]
bq. But just for my future reference - do we not introduce any new test
dependencies in oak ? or is there some RTC process that I bypassed and needed
to follow before introducing one ?
No, adding new test dependencies is fine. We just should do it consciously and
coordinated. My preference would be not to add new dependencies just because
somebody liked a particular library. In the end, the whole team needs to be
able to support the code in the future.
was (Author: reschke):
{quote}[~reschke] - Just find it a bit more fluent with assertions - makes the
test code a bit more clean to look at and I used the version that I had used
before to make sure it works as I expect . But I can remove it - I also need to
refactor the test to use the LogCustomizer as you suggested - so will remove
this dependency also along with that.
{quote}
Done that already, and I don't thik the test is harder to read. Will apply
patch soonish.
bq. But just for my future reference - do we not introduce any new test
dependencies in oak ? or is there some RTC process that I bypassed and needed
to follow before introducing one ?
No, adding new test dependencies is fine. We just should do it consciously and
coordinated. My preference would be not to add new dependencies just because
somebody liked a particular library. In the end, the whole team needs to be
able to support the code in the future.
> Debug logging when two or more indices have same or very close cost amounts
> doesn't work in case both indices belong to the same type of Query Index
>
>
> Key: OAK-8035
> URL: https://issues.apache.org/jira/browse/OAK-8035
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: query
>Reporter: Nitin Gupta
>Assignee: Julian Reschke
>Priority: Major
> Fix For: 1.12, 1.11.0
>
> Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch
>
>
> Steps to reproduce -
> # You would need two index definitions of same query index type , let's say
> fullText index , having similar cost evaluations.
> # Now while the index plan is evaluated you will notice that there should
> have been a debug log saying that selected index has similar cost - this
> serves as an important warning to either change the index definitions or the
> query
> ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)]
>
> However If we have a look at this code block here -
> [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041]
> , there is no comparison for near best costs for index plans having same
> query index type .
>
> cc : [~teofili]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[
https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771825#comment-16771825
]
Julian Reschke edited comment on OAK-8035 at 2/19/19 11:18 AM:
---
{quote}[~reschke] - Just find it a bit more fluent with assertions - makes the
test code a bit more clean to look at and I used the version that I had used
before to make sure it works as I expect . But I can remove it - I also need to
refactor the test to use the LogCustomizer as you suggested - so will remove
this dependency also along with that.
{quote}
Done that already, and I don't thik the test is harder to read. See
[r1853872|http://svn.apache.org/r1853872].
bq. But just for my future reference - do we not introduce any new test
dependencies in oak ? or is there some RTC process that I bypassed and needed
to follow before introducing one ?
No, adding new test dependencies is fine. We just should do it consciously and
coordinated. My preference would be not to add new dependencies just because
somebody liked a particular library. In the end, the whole team needs to be
able to support the code in the future.
was (Author: reschke):
{quote}[~reschke] - Just find it a bit more fluent with assertions - makes the
test code a bit more clean to look at and I used the version that I had used
before to make sure it works as I expect . But I can remove it - I also need to
refactor the test to use the LogCustomizer as you suggested - so will remove
this dependency also along with that.
{quote}
Done that already, and I don't thik the test is harder to read. See
[r1853872|http://svn.apache.org/r1853872]
bq. But just for my future reference - do we not introduce any new test
dependencies in oak ? or is there some RTC process that I bypassed and needed
to follow before introducing one ?
No, adding new test dependencies is fine. We just should do it consciously and
coordinated. My preference would be not to add new dependencies just because
somebody liked a particular library. In the end, the whole team needs to be
able to support the code in the future.
> Debug logging when two or more indices have same or very close cost amounts
> doesn't work in case both indices belong to the same type of Query Index
>
>
> Key: OAK-8035
> URL: https://issues.apache.org/jira/browse/OAK-8035
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: query
>Reporter: Nitin Gupta
>Assignee: Julian Reschke
>Priority: Major
> Fix For: 1.12, 1.11.0
>
> Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch
>
>
> Steps to reproduce -
> # You would need two index definitions of same query index type , let's say
> fullText index , having similar cost evaluations.
> # Now while the index plan is evaluated you will notice that there should
> have been a debug log saying that selected index has similar cost - this
> serves as an important warning to either change the index definitions or the
> query
> ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)]
>
> However If we have a look at this code block here -
> [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041]
> , there is no comparison for near best costs for index plans having same
> query index type .
>
> cc : [~teofili]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[ https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771857#comment-16771857 ] Nitin Gupta edited comment on OAK-8035 at 2/19/19 11:59 AM: [~reschke] - agreed . I was just about to add this patch here [^OAK-8035_4.patch] for the changes as per your suggestions - but you beat me to it . Thanks for the help :) was (Author: nitigup): [~reschke] - agreed . I was just about to add this patch here [^OAK-8035_4.patch] for the changes as per your suggestions - but you beat me to it :) > Debug logging when two or more indices have same or very close cost amounts > doesn't work in case both indices belong to the same type of Query Index > > > Key: OAK-8035 > URL: https://issues.apache.org/jira/browse/OAK-8035 > Project: Jackrabbit Oak > Issue Type: Bug > Components: query >Reporter: Nitin Gupta >Assignee: Julian Reschke >Priority: Major > Fix For: 1.12, 1.11.0 > > Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch, > OAK-8035_4.patch > > > Steps to reproduce - > # You would need two index definitions of same query index type , let's say > fullText index , having similar cost evaluations. > # Now while the index plan is evaluated you will notice that there should > have been a debug log saying that selected index has similar cost - this > serves as an important warning to either change the index definitions or the > query > ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)] > > However If we have a look at this code block here - > [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041] > , there is no comparison for near best costs for index plans having same > query index type . > > cc : [~teofili] -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Comment Edited] (OAK-8035) Debug logging when two or more indices have same or very close cost amounts doesn't work in case both indices belong to the same type of Query Index
[
https://issues.apache.org/jira/browse/OAK-8035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771857#comment-16771857
]
Nitin Gupta edited comment on OAK-8035 at 2/19/19 12:03 PM:
[~reschke] - agreed - will keep this in mind in future .
{noformat}
No, adding new test dependencies is fine. We just should do it consciously and
coordinated. My preference would be not to add new dependencies just because
somebody liked a particular library. In the end, the whole team needs to be
able to support the code in the future{noformat}
I was just about to add this patch here [^OAK-8035_4.patch] for the changes as
per your suggestions - but you beat me to it . Thanks for the help :)
was (Author: nitigup):
[~reschke] - agreed .
I was just about to add this patch here [^OAK-8035_4.patch] for the changes as
per your suggestions - but you beat me to it . Thanks for the help :)
> Debug logging when two or more indices have same or very close cost amounts
> doesn't work in case both indices belong to the same type of Query Index
>
>
> Key: OAK-8035
> URL: https://issues.apache.org/jira/browse/OAK-8035
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: query
>Reporter: Nitin Gupta
>Assignee: Julian Reschke
>Priority: Major
> Fix For: 1.12, 1.11.0
>
> Attachments: OAK-8035.patch, OAK-8035_2.patch, OAK-8035_3.patch,
> OAK-8035_4.patch
>
>
> Steps to reproduce -
> # You would need two index definitions of same query index type , let's say
> fullText index , having similar cost evaluations.
> # Now while the index plan is evaluated you will notice that there should
> have been a debug log saying that selected index has similar cost - this
> serves as an important warning to either change the index definitions or the
> query
> ([https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1069#L1072)]
>
> However If we have a look at this code block here -
> [https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/query/QueryImpl.java#L1017#L1041]
> , there is no comparison for near best costs for index plans having same
> query index type .
>
> cc : [~teofili]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
