[ 
https://issues.apache.org/jira/browse/IGNITE-14783?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17351795#comment-17351795
 ] 

Sergei Ryzhov edited comment on IGNITE-14783 at 5/26/21, 2:05 PM:
------------------------------------------------------------------

*SystemViewComputeJobTest.testCancelComputeTask*
when job cancel, job.finishTime more than 0
the test flaky because job.finishTime does not always have time to update

*IoStatisticsSelfTest.testPersistentIOGlobalStat*
in the PageMemoryImpl.acquirePage(int grpId, long pageId, IoStatisticsHolder 
statHolder, boolean restore, @nullable AtomicBoolean pageAllocated)

{code:java}
long relPtr = seg.loadedPages.get(
                grpId,
                PageIdUtils.effectivePageId(pageId),
                seg.partGeneration(grpId, partId),
                INVALID_REL_PTR,
                INVALID_REL_PTR
            );

// The page is loaded to the memory.
if (relPtr != INVALID_REL_PTR) {
    long absPtr = seg.absolute(relPtr);

    seg.acquirePage(absPtr);

    seg.pageReplacementPolicy.onHit(relPtr);

    statHolder.trackLogicalRead(absPtr + PAGE_OVERHEAD);

    return absPtr;
}
{code}

so we don't call IoStatisticsHolderCache.trackPhysicalAndLogicalRead

the call to IoStatisticsHolderCache.trackPhysicalAndLogicalRead occurs when 
reading from the disk, it does not come as long as the data is in memory

one of the solutions, to increase the amount of loaded data


was (Author: ryzhovsv):
*SystemViewComputeJobTest.testCancelComputeTask*
when job cancel, job.finishTime more than 0
the test flaky because job.finishTime does not always have time to update

*IoStatisticsSelfTest.testPersistentIOGlobalStat*
in the PageMemoryImpl.acquirePage(int grpId, long pageId, IoStatisticsHolder 
statHolder, boolean restore, @nullable AtomicBoolean pageAllocated)

{code:java}
long relPtr = seg.loadedPages.get(
                grpId,
                PageIdUtils.effectivePageId(pageId),
                seg.partGeneration(grpId, partId),
                INVALID_REL_PTR,
                INVALID_REL_PTR
            );

// The page is loaded to the memory.
if (relPtr != INVALID_REL_PTR) {
    long absPtr = seg.absolute(relPtr);

    seg.acquirePage(absPtr);

    seg.pageReplacementPolicy.onHit(relPtr);

    statHolder.trackLogicalRead(absPtr + PAGE_OVERHEAD);

    return absPtr;
}
{code}

so we don't call IoStatisticsHolderCache.trackPhysicalAndLogicalRead

one of the solutions, to increase the amount of loaded data

> Fix tests fail: IgniteCacheTestSuite9:SystemViewComputeJobTest, 
> IoStatisticsSelfTest
> ------------------------------------------------------------------------------------
>
>                 Key: IGNITE-14783
>                 URL: https://issues.apache.org/jira/browse/IGNITE-14783
>             Project: Ignite
>          Issue Type: Test
>            Reporter: Sergei Ryzhov
>            Assignee: Sergei Ryzhov
>            Priority: Minor
>          Time Spent: 0.5h
>  Remaining Estimate: 0h
>
> Fix tests fail:
> IgniteCacheTestSuite9:
>  - SystemViewComputeJobTest.testCancelComputeTask
>  - IoStatisticsSelfTest.testPersistentIOGlobalStat



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to