Open JasonFengJ9 opened 4 years ago
fyi @hangshao0
JVMSHRC226E Error opening shared class cache file
JVMSHRC336E Port layer error code = -108
The test code wants to create a test cache, but it can't, leading to the following failure. I am not aware of any recent changes to the test or VM code.
Does -108 give a clue as to why it couldn't create the cache?
Platform error message: No such file or directory
Looking at the core more closely:
!sh_oscache 0x000001000320E680
SH_OSCache at 0x1000320e680 {
Fields for SH_OSCache:
...
0x60: UDATA _createFlags = 0x0000000000000004 (4)
The create flag is J9SH_OSCACHE_OPEXIST_STATS (==4), which attempts to open an existing cache.
I see the following before the portlib error:
openTestCache: Opening test cache
openTestCache: Done opening test cache
What happened is more likely: The test cache was successfully created, but later it was deleted somehow. We try to open the cache again with J9SH_OSCACHE_OPEXIST_STATS, finding it is gone, seeing portlib error -108.
I guess this might be similar to #8511, #8883, #9021. Caches are deleted/created unexpectedly, seems that we are seeing this on PPCLE only now.
Is is a persistent or non-persistent cache?
I wonder if people are using the machines without reserving them, and caches get removed as a side affect of manually running testing.
@JasonFengJ9 it would be good to record the machine name.
I filled in the machine name where available. #8883, #9021 are the same machine, this one is different.
@jdekonin is there any kind of machine sharing going on? "Caches are deleted/created unexpectedly" internally on ppcle.
Is is a persistent or non-persistent cache?
I see:
Running tests with cacheType: 1(J9PORT_SHR_CACHE_TYPE_PERSISTENT)
It is persistent cache.
There shouldn't be resource sharing no. I only see rhel7lert1
mentioned. What would be another machine name and I can say if its even on the same kvm host.
@jdekonin the other machine is cent7p8j91
ah...those might have been. cent7p8j91 has since been deleted as the reason for its origin was not known. It was the same host though.
Failure link
From an internal build
Test_openjdk11_j9_sanity.functional_ppc64le_linux_Nightly/341/
rhel7lert1Optional info
Failure output (captured from console output)
In addition, there was a GPF at end of the test.
To rebuild the failed tests in=https://hyc-runtimes-jenkins.swg-devops.com/job/Grinder, use the following links: https://hyc-runtimes-jenkins.swg-devops.com/job/Grinder/parambuild/?JDK_VERSION=11&JDK_IMPL=openj9&BUILD_LIST=functional&PLATFORM=ppc64le_linux&TARGET=shrtest_linux_0