Open sxa opened 1 year ago
Failures occurred on the s390x RHEL8 system too ... Will assign myself so I can discuss with the Red Hat internal team.
Upstream issue: https://bugs.openjdk.org/browse/JDK-8295343
@smlambert What would you propose as the next action here?
Exclude the tests in our problem lists under https://bugs.openjdk.org/browse/JDK-8295343 (and no we currently do not support by OS sub-flavours and supporting such granular exclusions is currently not in plan, unless somehow jtreg already offers that feature), and try to see an expedited fix to JDK-8295343 to be able to re-enable the tests quickly.
See also https://ci.adoptium.net/job/Test_openjdk8_hs_extended.openjdk_x86-64_linux_testList_2/86/testReport/ and https://ci.adoptium.net/job/Test_openjdk8_hs_extended.openjdk_x86-64_linux_testList_0/82/testReport/ which ran on test-equinix_esxi-ubuntu2204-x64-1 and test-equinix_esxi-ubuntu2204-x64-2, suggesting that Ubuntu 22.04 may be subject to the same issues as they both failed the following in the jdk_security3
suites:
New grinders to determine problematic tests
RHEL8/x64 failures (JDK8):
SSLHandshakeException: Received fatal alert: certificate_unknown
RSA key must be at least 1023 bits
RSA key must be at least 1023 bits
RSA key must be at least 1023 bits
PKCS11Exception: CKR_ARGUMENTS_BAD
Could not create RSA public key
Could not create RSA public key
Exception: Keys not equal
Exception: Keys not equal
exit code 1
On test-docker-ubuntu2204-x64-1 (JDK20):
SSLHandshakeException: Received fatal alert: certificate_unknown
Exception: Keys not equal
Exception: Keys not equal
On test-equinix_esxi-ubuntu2204-x64-1 (JDK8):
Ubuntu 20.04: https://ci.adoptium.net/job/Grinder/6905/ (Same three failures as Ubuntu 22.04)
RHEL8/x64 JDK20: https://ci.adoptium.net/job/Grinder/6909/console - PASSED
Ubuntu 22.04 x64 JDK20: All passed: https://ci.adoptium.net/job/Grinder/6904/
(I was assuming we'd have a nice simple list of things to exclude ... :-) )
On RHEL8/x64: JDK8 (14 failures) JDK11 (16 failures) JDK17 (12 failures) JDK19 (12 failures) JDK20 Nightly (0 failures!) (Note - failures can be /2 since each one is there twice for the _0
and _1
suites)
Test | JDK8 | JDK11 | JDK17 | JDK19 | JDK20 |
---|---|---|---|---|---|
pkcs11/Signature/ByteBuffers [*] | ❌ | ❌ | ✅ | ✅ | ✅ |
pkcs11/Signature/ReinitSignature [*] | ❌ | ❌ | ✅ | ✅ | ✅ |
pkcs11/Signature/TestRSAKeyLength [*] | ❌ | ❌ | ❌ | ❌ | n/a |
pkcs11/fips/TestTLS12 | ❌ | ✅ | n/a | n/a | n/a |
tools/autotest.sh | ❌ | ❌ | n/a | n/a | n/a |
pkcs11/rsa/TestKeyFactory [+] | ❌ | ❌ | ❌ | ❌ | ✅ |
pkcs11/rsa/TestSignatures | ❌ | ❌ | ❌ | ❌ | ✅ |
pkcs11/KeyStore/Basic.java | ✅ | ❌ | ❌ | ❌ | n/a |
pkcs11/KeyPair/TestKeyPairGenerator | n/a | ❌ | ❌ | ❌ | n/a |
tools/NssTest | n/a | n/a | ❌ | ❌ | n/a |
lib/cacerts/VerifyCACerts | n/a | n/a | n/a | ❌ | n/a |
X509TrustManager/Distrust.java | n/a | n/a | n/a | ❌ | n/a |
[*] - the top 3 in this table are the ones definitely related to the enforced minimum key lengths in RHEL8, and suggests a fix may have gone into to 17+ [+] - Note that while sun/security/pkcs11/rsa/TestKeyFactory.java.TestKeyFactory is a failure, sun/security/pkcs11/rsa/TestKeyFactory passed
Need to determine correct actions in our exclude list or upstream in order to help this to pass consistently and run the appropriate tests.
@judovana Can you confirm whether you've seen these on your systems, and if you've excluded them in line with the table above?
I had checked all our rhel8 jobs, and the tests you are reffering are run, and passing. More details sent offlist.
Some new tests (Since the ones in the table above were from about 18 months ago)
JDK | OS | arch | suite | Results |
---|---|---|---|---|
jdk8u | RHEL8 | x64 | jdk_security3_0 | 1 failure sun/security/pkcs11/fips/TestTLS12.java.TestTLS12 |
jdk8u | UBI9 | x64 | jdk_security3_0 | 7 failures (*) |
jdk8u | UBI9 | x64 | jdk_security3 | Same 7 failures as above (each duplicated twice) |
jdk8u | UBI9 | arm64 | jdk_security3 | 1 failure - autotest.sh |
jdk8u | UBI8 | arm64 | jdk_security3 | PASS |
jdk8u | UBI9 | arm64 | jdk_security3_0 | 1 failure keytool/autotest.sh |
jdk8u | CS9 | arm64 | jdk_security3 | 2 failures autotest.sh and https://ci.adoptium.net/job/Grinder/11566/testReport/javax_net_ssl_ServerName_SSLEngineExplorerMatchedSNI/java/SSLEngineExplorerMatchedSNI/ javax.net.ssl.SSLProtocolException: Input record too big: max = 16709 len = 42359 . Running with 100 iterations SNI passed 100/100 |
jdk8u(RH) | RHEL8 | x64 | jdk_security3 | 3 failures autotest.sh , standard.sh and TestTLS12 |
jdk8u(RH) | UBI9 | x64 | jdk_security3 | 8 failures 5 Secmod , standard.sh autotest.shand Login.sh` |
jdk8u(Zulu) | RHEL8 | x64 | jdk_security3_0 | Three failures |
jdk8u(Zulu) | UBI9 | x64 | jdk_security3_0 | 10 failures |
jdk8u(Zulu) | UBI9 | x64 | jdk_security3_0 | 10 failures |
jdk8u(Bish) | RHEL8 | x64 | jdk_security3_0 | 5 failures |
jdk8u(D'well) | RHEL8 | x64 | jdk_security3_0 | 8 failures |
jdk8u(Semeru) | RHEL8 | x64 | jdk_security3_0 | 1 failure SNI test (failed on Temurin on CS9) |
jdk8u(Semeru) | UBI9 | x64 | jdk_security3_0 | 5 failures Three Secmod issues, Login.sh and autotest.sh |
jdk11u | RHEL8 | x64 | jdk_security3_0 | PASS |
jdk17u | RHEL8 | x64 | jdk_security3_0 | 4 failures |
jdk11u | UBI9 | x64 | jdk_security3_0 | PASS |
jdk17u | UBI9 | x64 | jdk_security3_0 | 4 failures - same as on RHEL8 |
jdk21u | RHEL8 | x64 | jdk_security3_0 | 4 failures - same as jdk17u |
jdk21u | UBI9 | x64 | jdk_security3_0 | 4 failures - same as jdk17u |
(*) Here are the seven failures on UBI9: sun/security/pkcs11/Provider/Login.sh.Login 1.4 sec 1 sun/security/pkcs11/Secmod/AddPrivateKey.java.AddPrivateKey 0.27 sec 1 sun/security/pkcs11/Secmod/GetPrivateKey.java.GetPrivateKey 0.26 sec 1 sun/security/pkcs11/Secmod/JksSetPrivateKey.java.JksSetPrivateKey 0.2 sec 1 sun/security/pkcs11/Secmod/LoadKeystore.java.LoadKeystore 0.21 sec 1 sun/security/pkcs11/Secmod/TrustAnchors.java.TrustAnchors 0.23 sec 1 sun/security/tools/keytool/autotest.sh.autotest 2.4 sec 2
Noting that when trying with the Red Hat jdk8u build there was one additional failure:
Zulu showed three failures on RHEL8:
And zulu had 10 on UBI9:
JDK17u/21u failures on RHEL8/UBI9:
Summary of the above (unless otherwise stated these comparisons are on jdk8u/x64)
keytool/autotest
failure occurs almost everywhere (except Temurin JDK8/RHEL8, although it did fail with the RH build)sun/security/pkcs11/Secmod
tests which faili with libsoftokn3 version not found, set to 0.0: /usr/lib64/libsoftokn3.so
DHEKeySigning
which gives Expected to generate ServerHello series messages of 1387 bytes, but not 1643
, CacertsLimit
(possibly a different cacerts bundle) and SSLExceptionForIOIssue
(A timeout issue)Builds without our custom CA Certs bundle for feeding to the upstream parameters in Grinder:
Running on UBI9 with the jdk21u build above at https://ci.adoptium.net/job/Grinder/11651/console (May take a while since there are a lot of jobs queued up)
Running on UBI9 with the jdk21u build above at https://ci.adoptium.net/job/Grinder/11651/console (May take a while since there are a lot of jobs queued up)
Showed the same four errors so it seems unrelated our custom cacerts.
Running with Red Hat jdk21u on UBI9: https://ci.adoptium.net/job/Grinder/11717 - Failed the same four tests.
Please set the title to indicate the test name and machine name where known.
To make it easy for the infrastructure team to repeat and diagnose, please answer the following questions:
TARGET
ofjdk_security3_0
andjdk_security3_1
Test_
job on https://ci.adoptopenjdk.net which showed the failure: https://ci.adoptopenjdk.net/job/Test_openjdk17_hs_extended.openjdk_x86-64_linux/102/testReport/Any other details: Split out from #2030 as that issue is resolved See also https://github.com/adoptium/infrastructure/issues/2360 and https://github.com/adoptium/infrastructure/issues/1829 which are related to issues on this RHEL8 system.