daos icon indicating copy to clipboard operation
daos copied to clipboard
verified publisher icon daos-stack

DAOS-14149 client: add compatible mode for libpil4dfs

Open wiliamhuang opened this issue 2 years ago • 10 comments

Test-tag: pil4dfs

avoid using fake fd to provide better compatibility with degraded performance in open, openat, and opendir etc. "D_IL_COMPATIBLE=1" or "D_IL_COMPATIBLE=true" turns on compatible mode.

Required-githooks: true

Before requesting gatekeeper:

  • [ ] Two review approvals and any prior change requests have been resolved.
  • [ ] Testing is complete and all tests passed or there is a reason documented in the PR why it should be force landed and forced-landing tag is set.
  • [ ] Features: (or Test-tag*) commit pragma was used or there is a reason documented that there are no appropriate tags for this PR.
  • [ ] Commit messages follows the guidelines outlined here.
  • [ ] Any tests skipped by the ticket being addressed have been run and passed in the PR.

Gatekeeper:

  • [ ] You are the appropriate gatekeeper to be landing the patch.
  • [ ] The PR has 2 reviews by people familiar with the code, including appropriate watchers.
  • [ ] Githooks were used. If not, request that user install them and check copyright dates.
  • [ ] Checkpatch issues are resolved. Pay particular attention to ones that will show up on future PRs.
  • [ ] All builds have passed. Check non-required builds for any new compiler warnings.
  • [ ] Sufficient testing is done. Check feature pragmas and test tags and that tests skipped for the ticket are run and now pass with the changes.
  • [ ] If applicable, the PR has addressed any potential version compatibility issues.
  • [ ] Check the target branch. If it is master branch, should the PR go to a feature branch? If it is a release branch, does it have merge approval in the JIRA ticket.
  • [ ] Extra checks if forced landing is requested
    • [ ] Review comments are sufficiently resolved, particularly by prior reviewers that requested changes.
    • [ ] No new NLT or valgrind warnings. Check the classic view.
    • [ ] Quick-build or Quick-functional is not used.
  • [ ] Fix the commit message upon landing. Check the standard here. Edit it to create a single commit. If necessary, ask submitter for a new summary.

wiliamhuang avatar Nov 05 '23 04:11 wiliamhuang

Bug-tracker data: Ticket title is 'To implement a compatible mode in libpil4dfs for better compatibility' Status is 'In Progress' Labels: 'intercept_lib' https://daosio.atlassian.net/browse/DAOS-14149

github-actions[bot] avatar Nov 05 '23 04:11 github-actions[bot]

Test stage Unit Test bdev on EL 8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/5/display/redirect

daosbuild1 avatar Nov 14 '23 05:11 daosbuild1

Test stage NLT on EL 8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/5/display/redirect

daosbuild1 avatar Nov 14 '23 05:11 daosbuild1

We may need a separate ticket and PR for more tests.

wiliamhuang avatar Nov 17 '23 14:11 wiliamhuang

Test stage Functional Hardware Medium Verbs Provider completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/7/execution/node/1511/log

daosbuild1 avatar Nov 17 '23 17:11 daosbuild1

Ticket title is 'To implement a compatible mode in libpil4dfs for better compatibility' Status is 'In Review' Labels: 'intercept_lib' https://daosio.atlassian.net/browse/DAOS-14149

github-actions[bot] avatar Mar 17 '24 04:03 github-actions[bot]

Test stage Build RPM on EL 9 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/8/execution/node/350/log

daosbuild1 avatar Mar 17 '24 04:03 daosbuild1

Test stage Build RPM on EL 8 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/8/execution/node/327/log

daosbuild1 avatar Mar 17 '24 04:03 daosbuild1

Test stage Build DEB on Ubuntu 20.04 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/8/execution/node/296/log

daosbuild1 avatar Mar 17 '24 04:03 daosbuild1

Test stage Build RPM on Leap 15.5 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/8/execution/node/353/log

daosbuild1 avatar Mar 17 '24 04:03 daosbuild1

Test stage Functional Hardware Medium Verbs Provider completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/10/execution/node/1541/log

daosbuild1 avatar Mar 18 '24 14:03 daosbuild1

Test stage Functional Hardware Medium Verbs Provider completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/14/execution/node/1498/log

daosbuild1 avatar Mar 29 '24 02:03 daosbuild1

For this PR whilst it's still work-in-progress I suggest that you should modify https://github.com/daos-stack/daos/blob/master/src/tests/ftest/dfuse/daos_build.py#L206 to do a "git checkout lei/DAOS-14149" so that the is trying to build itself for the daos_build test. This obviously can't land like this but will be more accurate in what it's trying to test.

You could probably also remove this line https://github.com/daos-stack/daos/blob/master/site_scons/prereq_tools/base.py#L1268 I've tried it locally and it doesn't appear to be required, it might help with the recursion in daos_init although I suspect it would still be an issue.

I also still think you should add any environment variables that pil4dfs requires to https://github.com/daos-stack/daos/blob/master/SConstruct#L365-L366 to ensure they're correctly propagates as well.

ashleypittman avatar Apr 09 '24 20:04 ashleypittman

For this PR whilst it's still work-in-progress I suggest that you should modify https://github.com/daos-stack/daos/blob/master/src/tests/ftest/dfuse/daos_build.py#L206 to do a "git checkout lei/DAOS-14149" so that the is trying to build itself for the daos_build test. This obviously can't land like this but will be more accurate in what it's trying to test.

You could probably also remove this line https://github.com/daos-stack/daos/blob/master/site_scons/prereq_tools/base.py#L1268 I've tried it locally and it doesn't appear to be required, it might help with the recursion in daos_init although I suspect it would still be an issue.

I also still think you should add any environment variables that pil4dfs requires to https://github.com/daos-stack/daos/blob/master/SConstruct#L365-L366 to ensure they're correctly propagates as well.

@ashleypittman Thank you! I made the changes as you suggested.

wiliamhuang avatar Apr 09 '24 20:04 wiliamhuang

Test stage Functional Hardware Medium completed with status UNSTABLE. https://build.hpdd.intel.com/job/daos-stack/job/daos//view/change-requests/job/PR-13294/16/testReport/

daosbuild1 avatar Apr 10 '24 09:04 daosbuild1

@ashleypittman Could you please review? Thank you very much!

wiliamhuang avatar Apr 15 '24 18:04 wiliamhuang

Test stage Functional on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/22/execution/node/1199/log

daosbuild1 avatar Apr 17 '24 00:04 daosbuild1

Test stage Functional Hardware Medium completed with status UNSTABLE. https://build.hpdd.intel.com/job/daos-stack/job/daos//view/change-requests/job/PR-13294/23/testReport/

daosbuild1 avatar Apr 17 '24 17:04 daosbuild1

The failure in hardware medium (dfuse/find.py) was reported in https://daosio.atlassian.net/browse/DAOS-15616.

wiliamhuang avatar Apr 17 '24 21:04 wiliamhuang

@ashleypittman Do you have more feedback? Thank you very much!

wiliamhuang avatar Apr 18 '24 14:04 wiliamhuang

Looking at the test logs there are about 280k read calls to dfuse vs 2.8m read calls for the nocache version of the test, therefore this is intercepting approximately 90% of reads, is this around what you'd expect?

I've also tried this locally and it always fails in the same spot near the beginning no matter the settings I try, this may be something my end though so I'll keep testing with this. Cloning into '/tmp/dnt_dfuse_sukz4v1t/dfuse_mount.yxpi3q5j/24aa666b-1359-474a-b247-51c7dd2925c7/daos/src/rdb/raft'... fatal: cannot chdir to '../../../src/rdb/raft': Invalid argument Unable to fetch in submodule path 'src/rdb/raft'; trying to directly fetch 12dbc1595fad8b570de1e336205f994f2b0e22f5: fatal: cannot chdir to '../../../src/rdb/raft': Invalid argument

Executables and shared libraries are normally accessed by /lib64/ld-linux-x86-64.so.2. read() performed by ld are not intercepted since dfuse can take advantage of caching in kernel. These might explain why ~10% read are not intercepted.

daos build test with libpil4dfs compatible mode works fine locally on a boro node. I recalled you also got issues in a simple command "rm" reported in ticket https://daosio.atlassian.net/browse/DAOS-15065. If you are fine to share your node with me, I could test on your node later. Thank you!

wiliamhuang avatar Apr 18 '24 17:04 wiliamhuang

Looking at the test logs there are about 280k read calls to dfuse vs 2.8m read calls for the nocache version of the test, therefore this is intercepting approximately 90% of reads, is this around what you'd expect? I've also tried this locally and it always fails in the same spot near the beginning no matter the settings I try, this may be something my end though so I'll keep testing with this. Cloning into '/tmp/dnt_dfuse_sukz4v1t/dfuse_mount.yxpi3q5j/24aa666b-1359-474a-b247-51c7dd2925c7/daos/src/rdb/raft'... fatal: cannot chdir to '../../../src/rdb/raft': Invalid argument Unable to fetch in submodule path 'src/rdb/raft'; trying to directly fetch 12dbc1595fad8b570de1e336205f994f2b0e22f5: fatal: cannot chdir to '../../../src/rdb/raft': Invalid argument

Executables and shared libraries are normally accessed by /lib64/ld-linux-x86-64.so.2. read() performed by ld are not intercepted since dfuse can take advantage of caching in kernel. These might explain why ~10% read are not intercepted.

Very few shared librarries or binaries are launched from dfuse are they? Only the built daos commands when the build wants to extract the man page. Perhaps some libraries are being used from dfuse still? Python binaries might come from dfuse but they're small and mostly scripts.

daos build test with libpil4dfs compatible mode works fine locally on a boro node. I recalled you also got issues in a simple command "rm" reported in ticket https://daosio.atlassian.net/browse/DAOS-15065. If you are fine to share your node with me, I could test on your node later. Thank you!

I will message you directly, lets look into this.

ashleypittman avatar Apr 18 '24 18:04 ashleypittman

Test stage Functional Hardware Medium completed with status UNSTABLE. https://build.hpdd.intel.com/job/daos-stack/job/daos//view/change-requests/job/PR-13294/29/testReport/

daosbuild1 avatar Apr 21 '24 20:04 daosbuild1

Test stage Functional on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/daos/view/change-requests/job/PR-13294/32/execution/node/1199/log

daosbuild1 avatar Apr 23 '24 01:04 daosbuild1

@ashleypittman Could you please review the changes since yesterday? Thank you! daos build test in CI already finished. Hardware tests would wait quite long time in queue.

wiliamhuang avatar Apr 23 '24 12:04 wiliamhuang

After merging dcache PR, daos build failed in local test. I am debugging. I will update this PR later today.

wiliamhuang avatar Apr 24 '24 16:04 wiliamhuang

Test stage Unit Test with memcheck on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/39/display/redirect

daosbuild1 avatar Apr 24 '24 17:04 daosbuild1

Test stage Unit Test bdev with memcheck on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/39/display/redirect

daosbuild1 avatar Apr 24 '24 17:04 daosbuild1

Test stage Unit Test on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/39/display/redirect

daosbuild1 avatar Apr 24 '24 17:04 daosbuild1

Test stage Unit Test bdev on EL 8.8 completed with status FAILURE. https://build.hpdd.intel.com/job/daos-stack/job/daos/job/PR-13294/39/display/redirect

daosbuild1 avatar Apr 24 '24 17:04 daosbuild1