This prevents getting into a situation where a container log cannot make
progress because we tried to rotate a file, got an error, and now the
file is closed. The next time we try to write a log entry it will try
and rotate again but error that the file is already closed.
I wonder if there is more we can do to beef up this rotation logic.
Found this issue while investigating missing logs with errors in the
docker daemon logs like:
```
Failed to log message for json-file: error closing file: close <file>:
file already closed
```
I'm not sure why the original rotation failed since the data was no
longer available.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
These dependencies have tagged releases, but we're currently
vendoring from master. Add a comment to give some "clue" what
version we're vendoring.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
full diff: https://github.com/sirupsen/logrus/compare/v1.4.2...v1.6.0
- Ability to DisableHTMLEscape when using the JSON formatter
- Support/fixes for go 1.14
- Many many bugfixes
- Add flag to disable quotes in TextFormatter
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Looks like this Dockerfile was not yet updated; possibly because
`go mod` issues when building buildx 0.3.1.
Given that the upstream repository uses vendoring, I:
- removed the `go mod download` step
- added `-mod=vendor` to use vendoring mode
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- add `set -e` to prevent linting warnings
- use `install` instead of `cp`
- use a subshell for the whole function, instead of starting one
inside it.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The containerd.installer was updated to also copy `containerd-shim-runc-v2`,
but `hack/make/binary-daemon` did not copy it to the bundles directory.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The empty `binary:`, `dynbinary:`, and `cross:` targets at the top were
only useful to provide the description for `make help`.
This patch splits the actual `binary`, `dynbinary`, and `cross` targets
to separate lines, introducing some slight duplicated code, but making
it slightly easier to read (and removing the "empty" targets).
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
If both a "versioned" and "non-versioned" version of a dependency
exists in vendor.conf, the "longest" path should be cloned last,
otherwise the versioned dependency may be deleted when cloning its
non-versioned variant.
For example, with the following vendor.conf:
github.com/coreos/go-systemd/v22 v22.0.0
github.com/coreos/go-systemd v17
Running vndr would;
1. recursively delete "vendor/src/github.com/coreos/go-systemd/v22:
2. start cloning "github.com/coreos/go-systemd/v22"
3. recursively delete "vendor/src/github.com/coreos/go-systemd"
4. start cloning "github.com/coreos/go-systemd"
This would lead to a conflicting situation; step 3. will remove
the dependency that was previously cloned (or in the process
of being cloned).
This patch sorts the dependencies by import-path, cloning the
shortest import paths first, which should prevent the race condition.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
full diff: https://github.com/moby/ipvs/compare/v1.0.0...v1.0.1
- Fix compatibility issue on older kernels (< 3.18) where the address
family attribute for destination servers do not exist
- Fix the stats attribute check when parsing destination addresses
- NetlinkSocketsTimeout should be a constant
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This blurb exists because we reused the same commit from an old patch,
and thus got the commit message with it. However the message is
confusing in this context.
It was suggested in review that we should remove the confusing message.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
This version avoids doing name lookups on creating tarball that
should be avoided in to not hit loading glibc shared libraries.
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
(cherry picked from commit aa6a9891b0)
Signed-off-by: Tibor Vass <tibor@docker.com>
Since we don't need the actual split values, instead of calling
`strings.Split`, which allocates new slices on each call, use
`strings.Index`.
This significantly reduces the allocations required when doing env value
replacements.
Additionally, pre-allocate the env var slice, even if we allocate a
little more than we need, it keeps us from having to do multiple
allocations while appending.
```
benchmark old ns/op new ns/op delta
BenchmarkReplaceOrAppendEnvValues/0-8 486 313 -35.60%
BenchmarkReplaceOrAppendEnvValues/100-8 10553 1535 -85.45%
BenchmarkReplaceOrAppendEnvValues/1000-8 94275 12758 -86.47%
BenchmarkReplaceOrAppendEnvValues/10000-8 1161268 129269 -88.87%
benchmark old allocs new allocs delta
BenchmarkReplaceOrAppendEnvValues/0-8 5 2 -60.00%
BenchmarkReplaceOrAppendEnvValues/100-8 110 0 -100.00%
BenchmarkReplaceOrAppendEnvValues/1000-8 1013 0 -100.00%
BenchmarkReplaceOrAppendEnvValues/10000-8 10022 0 -100.00%
benchmark old bytes new bytes delta
BenchmarkReplaceOrAppendEnvValues/0-8 192 24 -87.50%
BenchmarkReplaceOrAppendEnvValues/100-8 7360 0 -100.00%
BenchmarkReplaceOrAppendEnvValues/1000-8 64832 0 -100.00%
BenchmarkReplaceOrAppendEnvValues/10000-8 1146049 0 -100.00%
```
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
The "systemd" cgroup driver is always preferred over "cgroupfs" on
systemd-based hosts.
This commit does not affect cgroup v1 hosts.
Signed-off-by: Akihiro Suda <akihiro.suda.cz@hco.ntt.co.jp>
