Bug 1897248
| Summary: | performance search rate: nagle triggers high rate of setsocketopt | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 8 | Reporter: | thierry bordaz <tbordaz> |
| Component: | 389-ds-base | Assignee: | thierry bordaz <tbordaz> |
| Status: | CLOSED ERRATA | QA Contact: | RHDS QE <ds-qe-bugs> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 8.3 | CC: | bsmejkal, mreynolds, sgouvern |
| Target Milestone: | rc | Keywords: | Triaged |
| Target Release: | 8.0 | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | sync-to-jira | ||
| Fixed In Version: | 389-ds-1.4-8040020201216214810.866effaa | Doc Type: | Bug Fix |
| Doc Text: |
Cause: In order to optimize network traffic and avoid sending partial LDAP responses, 389-ds uses TCP_CORK socket option. This socket option is useless because LDAP server always send complete responses/results
Consequence: To use TCP_CORK, LDAP server does setsockopt syscall per resp/result. But the use of TCP_CORK is useless in case of 389-ds
Fix: set nagle=off by default
Result: reduce the number of setsockopt calls
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | 2021-05-18 15:45:43 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
This change has an impact on documentation: https://access.redhat.com/documentation/en-us/red_hat_directory_server/11/html/configuration_command_and_file_reference/core_server_configuration_reference#cnconfig-nsslapd_nagle ============================================================================================================ test session starts =============================================================================================================
platform linux -- Python 3.6.8, pytest-6.2.1, py-1.10.0, pluggy-0.13.1 -- /usr/bin/python3.6
cachedir: .pytest_cache
metadata: {'Python': '3.6.8', 'Platform': 'Linux-4.18.0-269.el8.x86_64-x86_64-with-redhat-8.4-Ootpa', 'Packages': {'pytest': '6.2.1', 'py': '1.10.0', 'pluggy': '0.13.1'}, 'Plugins': {'metadata': '1.11.0', 'html': '3.1.1', 'libfaketime': '0.1.2'}}
389-ds-base: 1.4.3.16-7.module+el8.4.0+9324+a82a8f71
nss: 3.53.1-17.el8_3
nspr: 4.25.0-2.el8_2
openldap: 2.4.46-16.el8
cyrus-sasl: 2.1.27-5.el8
FIPS: disabled
rootdir: /mnt/tests/rhds/tests/upstream/ds/dirsrvtests, configfile: pytest.ini
plugins: metadata-1.11.0, html-3.1.1, libfaketime-0.1.2
collected 11 items / 10 deselected / 1 selected
dirsrvtests/tests/suites/config/config_test.py::test_nagle_default_value PASSED [100%]
================================================================================================ 1 passed, 10 deselected in 8.69s ===========================================================================================================
Marking as Verified:Tested.
wrong build attached to the errata -> moving to ITM12 Correct build now attached to the errata : as per comment 3, marking as VERIFIED and moving back to ITM11 Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (389-ds:1.4 bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:1835 |
Description of problem: nagle is enabled by default. To prevent returned search entries/result to be stuck the server set/reset (setsocketopt) TCP_CORK. It can give a network benefit in term of transmitted overhead, but the cost is a big use of systemcall setsocketopt. The benefit of nagle should be limited by the fact that DS always write complete PDU (response or entries) Switching on/off nagle has no significant impact on throughput but gives benefit in terms of syscall tracing: perf trace -t <worker> -s NAGLE: off ns-slapd (108026), 1527476 events, 100.0% syscall calls total min avg max stddev (msec) (msec) (msec) (msec) (%) --------------- -------- --------- --------- --------- --------- ------ futex 328122 3059.967 0.000 0.009 360.607 19.06% poll 124478 854.697 0.002 0.007 100.142 20.29% sendto 124149 785.173 0.002 0.006 0.041 0.11% write 123627 483.753 0.002 0.004 0.080 0.10% recvfrom 62295 245.209 0.002 0.004 0.040 0.12% madvise 9 0.211 0.016 0.023 0.042 12.46% sched_yield 34 0.146 0.004 0.004 0.005 1.32% NAGLE: on ns-slapd (108026), 1784094 events, 100.0% syscall calls total min avg max stddev (msec) (msec) (msec) (msec) (%) --------------- -------- --------- --------- --------- --------- ------ futex 328704 3054.546 0.000 0.009 346.687 17.23% setsockopt 124702 656.760 0.001 0.005 0.034 0.14% poll 125437 655.709 0.002 0.005 100.141 15.28% sendto 124501 464.307 0.001 0.004 0.039 0.14% write 125026 462.438 0.001 0.004 0.031 0.10% recvfrom 62799 248.383 0.002 0.004 0.030 0.11% madvise 3 0.066 0.014 0.022 0.034 26.97% sched_yield 10 0.043 0.004 0.004 0.004 1.69% Version-Release number of selected component (if applicable): How reproducible: Will be described later Actual results: With searchrate, we can see as much setsockopt than the #write Expected results: See lower number of setsockopt