+1 Has been running stable with production traffic for 22 hours. Also, docs has been running stable for over 24 hours with an ASan built version of rc1.
Thank you Chris for working on this! On Thu, Aug 13, 2026 at 11:34 AM Bryan Call <[email protected]> wrote: > +1 > > Verified the 10.2.0-rc1 artifacts and ran the full test suite on two > machines, > plus a Fedora 45 container to confirm the two issues found in RC0 are > resolved. > > Built and tested on: > - OS: Fedora Linux 44 (Server Edition), x86_64 > - Compiler: gcc (GCC) 16.1.1 20260515 (Red Hat 16.1.1-2) > - CMake 4.3.0 > > Results: > - SHA512 checksum: PASS > - GPG signature: PASS (good signature, key > 894CE665B46B08390F3E6DB1A733550146C06766) > - Build (cmake --preset release): PASS > - Unit tests (ctest): PASS (90/90) > - Regression (traffic_server -R 3): PASS (215/215) > - AuTest (541 tests, 8 parallel workers): 519 passed, 2 failed, 38 skipped, > 0 exceptions > > Both RC0 problems are fixed. I rebuilt RC1 in a Fedora 45 (Rawhide, glibc > 2.43.9000) container: the build is clean and ctest is 84 of 84 with no > failures. On RC0 the same container failed the libswoc errno test, because > glibc has now assigned errno 134 to EFTYPE ("Inappropriate file type or > format") and the test assumed 134 had no message. The fastlz flz_maxcopy > signature change is also present in the RC1 tarball. > > The two AuTest failures are both pre-existing flaky tests, not defects in > this > candidate: > > - TSVConnFd passed three out of three times when re-run on its own. > > - bad_http_fmt fails intermittently because the test's own origin server > cannot bind its port, so the client sees a connection refused where the > gold > file expects a 200 response. The test reserves a port with > Test.GetTcpPort() and then hands the number to an external script that > binds > it later, which leaves a window for another process to take it. It is not > listed in serial_tests.txt, so it runs in the parallel pool where port > pressure is highest. It also failed on RC0, so it is not new here. No > Traffic Server code is involved in the failure. > > Unit tests and the regression suite passed identically on both machines. > > -Bryan > > On Aug 12, 2026, at 1:01 PM, Chris McFarlen <[email protected]> wrote: > > I've prepared a release for 10.2.0. This is the second release candidate > (rc1). Due to issue found in testing RC0, the prior vote did not pass. > Please retest and vote again. > > This is RC1 and includes the following commits: > > #13531 - Change libswoc test for undefined errno to a much higher number > #13532 - Fix signature on flz_maxcopy to avoid cast error > #13535 - Fix an intermittent cache unit test deadlock at exit > #13536 - cache: fix shm sizing on large-page Linux > #13537 - Share regex_remap rule sets > > ChangeLog: > > > https://github.com/apache/trafficserver/blob/10.2.x/CHANGELOG-10.2.0 > > The artifacts are available for download at: > > https://dist.apache.org/repos/dist/dev/trafficserver/10.2.0 > > SHA512 checksum: > > > dc32cb6da0fc0b4eb4dc0ac7c1bb465eebc3fa27ff5c8611b686958b295a548492b336f0b4356d1848999ada2988d926c4538f678c68b0ba38cf3252bef72119 > > This corresponds to git refs: > > Hash: b1df7334b39a878971c376da849e8da4984918b6 > Tag: 10.2.0-rc1 > > Which can be verified with the following command: > > $ git tag -v 10.2.0-rc1 > > All code signing keys are available here: > > https://downloads.apache.org/trafficserver/KEYS > > Make sure you refresh from a key server to get all relevant signatures > > Please test and cast your votes as early as possible. > > > -Chris > > > Sent with Proton Mail <https://proton.me/mail/home> secure email. > > > -- "Come to Me, all who are weary and heavy-laden, and I will give you rest. Take My yoke upon you and learn from Me, for I am gentle and humble in heart, and you will find rest for your souls. For My yoke is easy and My burden is light." ~ Matthew 11:28-30
