test: unflake async hooks statwatcher test by lundibundi · Pull Request #30362 · nodejs/node
added
async_hooks
labels
Nov 11, 2019Qard approved these changes Dec 21, 2019
Trott
added
the
author ready
label
Dec 21, 2019
Under high load 2 types of issues arise with this test. * filesystem calls gets queued even when the 'sync' is used which leads to async_hooks being called with the events of tmpdir clean or initial file write after clean. This is solved by counting all 'change' calls while making sure there is no dependency of StatWatcher's on one another and the expected changes are waited for. * some events are getting lost with the current clean->write->write_and_watch strategy. Specifically I observed the file size going from 0 to 5 entirely skipping 3 even though the write call was there (this happened reliably on -j128). So I've changed the strategy to avoid additional write considering this still tests the hooks correctly. This may indicate some sort of bug in async_hooks though I'm not sure. Closes: nodejs#21425
Trott pushed a commit that referenced this pull request
Dec 22, 2019
Under high load 2 types of issues arise with this test. * filesystem calls gets queued even when the 'sync' is used which leads to async_hooks being called with the events of tmpdir clean or initial file write after clean. This is solved by counting all 'change' calls while making sure there is no dependency of StatWatcher's on one another and the expected changes are waited for. * some events are getting lost with the current clean->write->write_and_watch strategy. Specifically I observed the file size going from 0 to 5 entirely skipping 3 even though the write call was there (this happened reliably on -j128). So I've changed the strategy to avoid additional write considering this still tests the hooks correctly. This may indicate some sort of bug in async_hooks though I'm not sure. Closes: #21425 PR-URL: #30362 Fixes: #21425 Reviewed-By: Stephen Belanger <admin@stephenbelanger.com> Reviewed-By: Rich Trott <rtrott@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Vladimir de Turckheim <vlad2t@hotmail.com>
BridgeAR pushed a commit that referenced this pull request
Jan 3, 2020
Under high load 2 types of issues arise with this test. * filesystem calls gets queued even when the 'sync' is used which leads to async_hooks being called with the events of tmpdir clean or initial file write after clean. This is solved by counting all 'change' calls while making sure there is no dependency of StatWatcher's on one another and the expected changes are waited for. * some events are getting lost with the current clean->write->write_and_watch strategy. Specifically I observed the file size going from 0 to 5 entirely skipping 3 even though the write call was there (this happened reliably on -j128). So I've changed the strategy to avoid additional write considering this still tests the hooks correctly. This may indicate some sort of bug in async_hooks though I'm not sure. Closes: #21425 PR-URL: #30362 Fixes: #21425 Reviewed-By: Stephen Belanger <admin@stephenbelanger.com> Reviewed-By: Rich Trott <rtrott@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Vladimir de Turckheim <vlad2t@hotmail.com>
targos pushed a commit that referenced this pull request
Jan 14, 2020
Under high load 2 types of issues arise with this test. * filesystem calls gets queued even when the 'sync' is used which leads to async_hooks being called with the events of tmpdir clean or initial file write after clean. This is solved by counting all 'change' calls while making sure there is no dependency of StatWatcher's on one another and the expected changes are waited for. * some events are getting lost with the current clean->write->write_and_watch strategy. Specifically I observed the file size going from 0 to 5 entirely skipping 3 even though the write call was there (this happened reliably on -j128). So I've changed the strategy to avoid additional write considering this still tests the hooks correctly. This may indicate some sort of bug in async_hooks though I'm not sure. Closes: #21425 PR-URL: #30362 Fixes: #21425 Reviewed-By: Stephen Belanger <admin@stephenbelanger.com> Reviewed-By: Rich Trott <rtrott@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Vladimir de Turckheim <vlad2t@hotmail.com>
BethGriggs pushed a commit that referenced this pull request
Feb 6, 2020
Under high load 2 types of issues arise with this test. * filesystem calls gets queued even when the 'sync' is used which leads to async_hooks being called with the events of tmpdir clean or initial file write after clean. This is solved by counting all 'change' calls while making sure there is no dependency of StatWatcher's on one another and the expected changes are waited for. * some events are getting lost with the current clean->write->write_and_watch strategy. Specifically I observed the file size going from 0 to 5 entirely skipping 3 even though the write call was there (this happened reliably on -j128). So I've changed the strategy to avoid additional write considering this still tests the hooks correctly. This may indicate some sort of bug in async_hooks though I'm not sure. Closes: #21425 PR-URL: #30362 Fixes: #21425 Reviewed-By: Stephen Belanger <admin@stephenbelanger.com> Reviewed-By: Rich Trott <rtrott@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Vladimir de Turckheim <vlad2t@hotmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters