On Thu, Jun 22, 2023 at 4:44 PM Wang Yugui <wangyugui@xxxxxxxxxxxx> wrote: > > Hi, > > > Hi all, > > > > When testing PostgreSQL, I found a performance degradation. After some > > investigation, it ultimately reached the attached simple C program and > > turned out that the performance degradation happens on only the xfs > > filesystem (doesn't happen on neither ext3 nor ext4). In short, the > > program alternately does two things to extend a file (1) call > > posix_fallocate() to extend by 8192 bytes and (2) call pwrite() to > > extend by 8192 bytes. If I do only either (1) or (2), the program is > > completed in 2 sec, but if I do (1) and (2) alternatively, it is > > completed in 90 sec. > > > > $ gcc -o test test.c > > $ time ./test test.1 1 > > total 200000 > > fallocate 200000 > > filewrite 0 > > > > real 0m1.305s > > user 0m0.050s > > sys 0m1.255s > > > > $ time ./test test.2 2 > > total 200000 > > fallocate 100000 > > filewrite 100000 > > > > real 1m29.222s > > user 0m0.139s > > sys 0m3.139s > > > > Why does it take so long in the latter case? and are there any > > workaround or configuration changes to deal with it? > > > > I test it on xfs linux 6.1.35 and 6.4-rc7 > > the result is almost same. > > $ time ./test test.1 1 > real 0m1.382s > > $ time ./test test.2 2 > real 0m9.262s > > linunx kernel version please. I test it on: $ uname -r 6.1.29-50.88.amzn2023.x86_64 and $ uname -r 4.18.0-372.9.1.el8.x86_64 Regards, -- Masahiko Sawada Amazon Web Services: https://aws.amazon.com
