The HWE kernel still has some instances of btrfs-transaction hanging for
2+ minutes but it recovers. Remember to regularly reset the sysctl for
reporting hung tasks or it'll just stop reporting new instances.
The first kernel used on the affected systems was -112 so I don't have
experience of the
It's a regression from xenial and took a few weeks to appear on bionic.
I tried using lock_page_killable() in __unmap_and_move() but as far as I
can tell kcompactd0 hasn't got stuck in the same way. Instead I now have
every btrfs_start_transaction suddenly blocking for all processes, which
is impo
I'm also getting kcompactd0 hanging on its own on another host, which
presumably blocks btrfs-transaction later on:
[606364.285752] INFO: task kcompactd0:27 blocked for more than 120 seconds.
[606364.285994] Not tainted 4.15.0-112-generic #113-Ubuntu
[606364.286129] "echo 0 > /proc/sys/kerne
Public bug reported:
Example 1:
[346911.187920] INFO: task kcompactd0:53 blocked for more than 120 seconds.
[346911.187938] Not tainted 4.15.0-112-generic #113-Ubuntu
[346911.187951] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables
this message.
[346911.187968] kcompactd0 D
4.4.131 includes "sctp: do not check port in sctp_inet6_cmp_addr" that
breaks combined AF_INET and AF_INET6 support in SCTP, it should be fixed
in 4.4.133 with "sctp: handle two v4 addrs comparison in
sctp_inet6_cmp_addr".
--
You received this bug notification because you are a member of Kernel
P
5 matches
Mail list logo
