Hi Khaled, I have finished testing the kernels - it took more time than I expected but the results are:
4.15.0-136-generic: no issues 4.15.0-137-generic: disk performance issue 4.15.0-139-generic: disk performance issue 4.15.0-140-generic: disk performance issue 4.15.0-142-generic: disk performance issue Each kernel was running for minimum 7 days. Those which have disk performance issue start to break after few days of normal workload. Here is an example of "iotop -a", when our disk operations are throttled: Actual DISK READ: 0.00 B/s | Actual DISK WRITE: 0.00 B/s TID PRIO USER DISK READ DISK WRITE SWAPIN IO> COMMAND 10908 be/4 barman 0.00 B 144.00 K 0.00 % 99.99 % pg_receivewal --dbname=dbname=replication host=XXXXXX replication=~o-password --directory=/data/barman/XXXXX/streaming --slot=barman 9494 be/4 barman 0.00 B 144.00 K 0.00 % 99.99 % rsync --server -logDtpre.iLsfxC . /data/barman/XXXXXX/incoming/0000000100003183000000ED 10907 be/4 barman 0.00 B 144.00 K 0.00 % 99.99 % pg_receivewal --dbname=dbname=replication host=XXXXXX replication~-password --directory=/data/barman/XXXXXXX/streaming --slot=barman 7148 be/4 barman 0.00 B 144.00 K 0.00 % 99.99 % rsync --server -logDtpre.iLsfxC . /data/barman/XXXXXX/incoming/000000010000D2E900000043 11144 be/4 barman 0.00 B 144.00 K 0.00 % 99.99 % pg_receivewal --dbname=dbname=replication host=XXXXX replication=~o-password --directory=/data/barman/XXXXX/streaming --slot=barman 17724 be/3 root 0.00 B 8.00 K 0.00 % 99.99 % auditd 11217 be/4 root 0.00 B 8.00 K 0.00 % 99.99 % syslog-ng -F 10003 be/4 barman 0.00 B 96.00 K 0.00 % 99.99 % gpg -e -r BarmanBackup 10219 be/4 barman 0.00 B 96.00 K 0.00 % 99.99 % gpg -e -r BarmanBackup 9247 be/4 barman 0.00 B 140.00 K 0.00 % 98.62 % rsync --server -logDtpre.iLsfxC . /data/barman/mel-attn-hostdb/incoming/0000000100004CE000000083 11391 be/4 barman 0.00 B 140.00 K 0.00 % 98.60 % pg_receivewal --dbname=dbname=replication host=XXXXXXX replication=~o-password --directory=/data/barmanXXXXXXX/streaming --slot=barman 9556 be/4 barman 0.00 B 92.00 K 0.00 % 98.53 % gpg -e -r BarmanBackup 4381 be/4 root 0.00 B 8.00 K 0.00 % 57.53 % sadc -F -L -S DISK 60 /var/log/sysstat 10959 be/4 root 0.00 B 4.00 K 0.00 % 26.29 % sadc -F -L -S DISK 60 /var/log/sysstat 15626 be/4 root 0.00 B 4.00 K 0.00 % 8.22 % sadc -F -L -S DISK 60 /var/log/sysstat So, I think it is ok to say that the problem was introduced with 4.15.0-137-generic. We will try to see the latest kernel in our next patching window. However, based on the fact that kernels 137,139,140,142 are not working correctly, could this be related to some tuning changes? In our case, we haven't done any changes to e.g. sysctl after patching. Many thanks for your help so far! -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1928744 Title: Disk IO very slow on kernel 4.15.0-142-generic To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1928744/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs