Commit graph

13 commits

Author SHA1 Message Date
Sebastiaan van Stijn
cf7a5be0f2
daemon: don't adjust oom-score if score is 0
This patch makes two changes if --oom-score-adj is set to 0

- do not adjust the oom-score-adjust cgroup for dockerd
- do not set the hard-coded -999 score for containerd if
  containerd is running as child process

Before this change:

oom-score-adj | dockerd       | containerd as child-process
--------------|---------------|----------------------------
-             | -500          | -500 (same as dockerd)
-100          | -100          | -100 (same as dockerd)
 0            |  0            | -999 (hard-coded default)

With this change:

oom-score-adj | dockerd       | containerd as child-process
--------------|---------------|----------------------------
-             | -500          | -500 (same as dockerd)
-100          | -100          | -100 (same as dockerd)
0             | not adjusted  | not adjusted

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2020-10-05 19:52:02 +02:00
Brian Goff
bef73d8b07 Wait for c8d process exit instead of polling API
In the containerd supervisor, instead of polling the healthcheck API
every 500 milliseconds we can just wait for the process to exit.

Signed-off-by: Brian Goff <cpuguy83@gmail.com>
2019-10-16 12:23:10 -07:00
Sebastiaan van Stijn
e554ab5589
Allow system.MkDirAll() to be used as drop-in for os.MkDirAll()
also renamed the non-windows variant of this file to be
consistent with other files in this package

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2019-08-08 15:05:49 +02:00
Sebastiaan van Stijn
c85fe2d224
Merge pull request #38522 from cpuguy83/fix_timers
Make sure timers are stopped after use.
2019-06-07 13:16:46 +02:00
Simão Reis
3134161be3 Fix nil pointer derefence on failure to connect to containerd
Signed-off-by: Simão Reis <smnrsti@gmail.com>
2019-01-30 12:41:54 -01:00
Brian Goff
eaad3ee3cf Make sure timers are stopped after use.
`time.After` keeps a timer running until the specified duration is
completed. It also allocates a new timer on each call. This can wind up
leaving lots of uneccessary timers running in the background that are
not needed and consume resources.

Instead of `time.After`, use `time.NewTimer` so the timer can actually
be stopped.
In some of these cases it's not a big deal since the duraiton is really
short, but in others it is much worse.

Signed-off-by: Brian Goff <cpuguy83@gmail.com>
2019-01-16 14:32:53 -08:00
Sebastiaan van Stijn
dd7799afd4
update containerd client and dependencies to v1.2.0
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2018-11-05 18:46:26 +01:00
Tibor Vass
34eede0296 Remove 'docker-' prefix for containerd and runc binaries
This allows to run the daemon in environments that have upstream containerd installed.

Signed-off-by: Tibor Vass <tibor@docker.com>
2018-09-24 21:49:03 +00:00
Sebastiaan van Stijn
06b9588c2d
Merge pull request #37759 from dmcgowan/fix-libcontainerd-startup-error
Add fail fast path when containerd fails on startup
2018-09-14 15:15:38 +02:00
Derek McGowan
ce0b0b72bc
Add fail fast path when containerd fails on startup
Prevents looping of startup errors such as containerd
not being found on the path.

Signed-off-by: Derek McGowan <derek@mcgstyle.net>
2018-09-13 17:34:52 -07:00
Derek McGowan
c3e3293843
Fix supervisor healthcheck throttling
Fix default case causing the throttling to not be used.
Ensure that nil client condition is handled.

Signed-off-by: Derek McGowan <derek@mcgstyle.net>
2018-09-04 11:00:28 -07:00
John Howard
5accd82634 Add containerd.WithTimeout(60*time.Second) to match old calls
Signed-off-by: John Howard <jhoward@microsoft.com>
2018-08-23 12:03:43 -07:00
Derek McGowan
dd2e19ebd5
libcontainerd: split client and supervisor
Adds a supervisor package for starting and monitoring containerd.
Separates grpc connection allowing access from daemon.

Signed-off-by: Derek McGowan <derek@mcgstyle.net>
2018-08-06 10:23:04 -07:00