Investigate flaky parallel/test-crypto-keygen #22978

Activity

1. Trott

   added

   flaky-testIssues and PRs involving tests that fail intermittently in CI.Issues and PRs involving tests that fail intermittently in CI.

   on Sep 20, 2018
2. Trott commented on Sep 20, 2018

   

   Trott

   on Sep 20, 2018

   MemberAuthor

   More actions

   Comment from @tniessen in #22660 (comment):

   @Trott I fear there might be a small chance (0.392156862745098% if my calculation is correct) of OpenSSL successfully decrypting a key it is not supposed to decrypt, but leading to an invalid key and thus causing this error message. Let me try to confirm that before fixing the test.
3. Trott commented on Sep 20, 2018

   

   Trott

   on Sep 20, 2018

   MemberAuthor

   More actions

   While stress-testing locally with -j 96 --repeat 192, I got a different error:

   === release test-crypto-keygen ===                    
   Path: parallel/test-crypto-keygen
   assert.js:662
       throw actual;
       ^
   
   Error: error:0D0680A8:asn1 encoding routines:asn1_check_tlen:wrong tag
       at Sign.sign (internal/crypto/sig.js:83:27)
       at testSignVerify (/Users/trott/io.js/test/parallel/test-crypto-keygen.js:43:42)
       at assert.throws (/Users/trott/io.js/test/parallel/test-crypto-keygen.js:234:7)
       at getActual (assert.js:574:5)
       at Function.throws (assert.js:692:24)
       at AsyncWrap.common.mustCall (/Users/trott/io.js/test/parallel/test-crypto-keygen.js:233:12)
       at AsyncWrap.<anonymous> (/Users/trott/io.js/test/common/index.js:350:15)
       at AsyncWrap.wrap.ondone (internal/crypto/keygen.js:42:14)
   Command: out/Release/node /Users/trott/io.js/test/parallel/test-crypto-keygen.js
   [02:08|% 100|+ 191|-   1]: Done 
4. Trott commented on Sep 20, 2018

   

   Trott

   on Sep 20, 2018

   MemberAuthor

   More actions

   Failed on a different platform in CI:

   https://ci.nodejs.org/job/node-test-commit-arm/18620/nodes=centos7-arm64-gcc6/console

   10:06:22 not ok 2388 parallel/test-crypto-keygen
   10:06:22   ---
   10:06:22   duration_ms: 240.71
   10:06:22   severity: fail
   10:06:22   exitcode: -15
   10:06:22   stack: |-
   10:06:22     timeout
   10:06:22     assert.js:662
   10:06:22         throw actual;
   10:06:22         ^
   10:06:22     
   10:06:22     Error: error:0D07207B:asn1 encoding routines:ASN1_get_object:header too long
   10:06:22         at Sign.sign (internal/crypto/sig.js:83:27)
   10:06:22         at testSignVerify (/home/iojs/build/workspace/node-test-commit-arm/nodes/centos7-arm64-gcc6/test/parallel/test-crypto-keygen.js:43:42)
   10:06:22         at assert.throws (/home/iojs/build/workspace/node-test-commit-arm/nodes/centos7-arm64-gcc6/test/parallel/test-crypto-keygen.js:171:7)
   10:06:22         at getActual (assert.js:574:5)
   10:06:22         at Function.throws (assert.js:692:24)
   10:06:22         at AsyncWrap.common.mustCall (/home/iojs/build/workspace/node-test-commit-arm/nodes/centos7-arm64-gcc6/test/parallel/test-crypto-keygen.js:170:12)
   10:06:22         at AsyncWrap.<anonymous> (/home/iojs/build/workspace/node-test-commit-arm/nodes/centos7-arm64-gcc6/test/common/index.js:350:15)
   10:06:22         at AsyncWrap.wrap.ondone (internal/crypto/keygen.js:42:14)
   10:06:22   ...
5. tniessen commented on Sep 20, 2018

   

   tniessen

   on Sep 20, 2018

   Member

   More actions

   @Trott This is most likely a randomly occuring problem with low probability (which obviously increases massively by being tested multiple times).

   Changing /bad decrypt/ to /bad decrypt|asn1 encoding routines/ should be enough if my assumption is correct, I'll try to confirm that.

   👍React with 👍1Reacted by Rich Trott
6. mscdex commented on Sep 20, 2018

   

   mscdex

   on Sep 20, 2018

   Contributor

   More actions

   There should be an opensslStack (or something similarly named) on the error message object that may help as well?
7. tniessen

   added a commit that references this issue on Sep 20, 2018

   test: fix flaky key pair generation test

   ...

   584903b
8. tniessen commented on Sep 20, 2018

   

   tniessen

   on Sep 20, 2018

   Member

   More actions

   Fixed in #22980, thanks for letting me know, @Trott. (Out of 25000 generated RSA keys, the test failed 98, which is close enough to the theoretical value to confirm my assumption about the decryption padding.)
9. tniessen

   closed this as completedin bad670con Sep 20, 2018
10. targos

    added a commit that references this issue on Sep 21, 2018

    test: fix flaky key pair generation test

    ...

    83278b2
11. sjarmak

    added a commit that references this issue on Jul 27, 2026

    test: fix flaky key pair generation test

    ...

    bad670c