If a previous git invocation crashes, it is possible that an orphaned
lock file (e.g. shallow.lock) is left on the filesystem. This previously
caused git-sync to crash loop because the lock file is never deleted.
This change adds a check in sanityCheckRepo for the existence of a git
lock file. If the git lock file exists at this stage, then initRepo will
re-initialize the repository.
Env-flags are "flags" that can only be set by env var (see caveat below).
All of the real flags have a corresponding env-flag (kind of, but not
really). The real goal was to deprecate `--password` but keep the env
var as a documented interface.
This does that (though --password still works) and updates the usage and
manual.
This allows some future work to follow the pattern. We do not register
every CLI flag as an env-flag because the help text would be
duplicative. This probably wants a wrapper API that allows declaring of
abstract flags, with CLI, env, or both sources.
Caveat:
ACTUALLY, these still have a flag, but the flag is specially named and
hidden. This makes testing a little easier where passing flags is
handled well but env vars is not.
Old way:
- ls-remote $ref $ref^{} and parse
- compare to current
- if changed, fetch
- update worktree
New way:
- fetch $ref
- compare to current
- if change, update worktree
The --ssh-key-file flag can be specified more than once and the
GITSYNC_SSH_KEY_FILE env var will be parsed like PATH.
Also adds e2e coverage for wrong-key and for multiple keys.
These fix the issues identified by the k/k inspired linter
configuration that we will be adding:
```
pkg/pid1/pid1.go:72:14: ST1005: error strings should not end with punctuation or newlines (stylecheck)
return 0, fmt.Errorf("unhandled exit status: 0x%x\n", status)
^
pkg/pid1/pid1.go:86:21: ST1005: error strings should not end with punctuation or newlines (stylecheck)
return false, 0, fmt.Errorf("wait4(): %w\n", err)
^
main.go:480:34: Error return value of `pflag.CommandLine.MarkDeprecated` is not checked (errcheck)
pflag.CommandLine.MarkDeprecated("branch", "use --ref instead")
^
main.go:483:34: Error return value of `pflag.CommandLine.MarkDeprecated` is not checked (errcheck)
pflag.CommandLine.MarkDeprecated("change-permissions", "use --group-write instead")
^
main.go:486:34: Error return value of `pflag.CommandLine.MarkDeprecated` is not checked (errcheck)
pflag.CommandLine.MarkDeprecated("dest", "use --link instead")
^
main.go:1897:16: Error return value of `io.WriteString` is not checked (errcheck)
io.WriteString(h, s)
^
main.go:555:2: ifElseChain: rewrite if-else to switch statement (gocritic)
if *flDeprecatedBranch != "" && (*flDeprecatedRev == "" || *flDeprecatedRev == "HEAD") {
^
```
I believe we print a new line anyway from e.g. envInt, which calls fmt.Fprintln.
Found by staticcheck:
main.go:127:17: error strings should not end with punctuation or newlines (ST1005)
main.go:157:13: error strings should not end with punctuation or newlines (ST1005)
main.go:187:13: error strings should not end with punctuation or newlines (ST1005)
main.go:217:13: error strings should not end with punctuation or newlines (ST1005)
Fix a few misc linter issues:
main_test.go:750:7: Error return value is not checked (errcheck)
touch(dirPath)
^
main_test.go:759:7: Error return value is not checked (errcheck)
touch(filePath)
^
main_test.go:768:7: Error return value is not checked (errcheck)
touch(newfilePath)
^
main.go:2089:3: S1023: redundant `return` statement (gosimple)
return
^
main.go:1935:13: S1039: unnecessary use of fmt.Sprintf (gosimple)
sshCmd += fmt.Sprintf(" -o StrictHostKeyChecking=no")
^
main.go:1152:33: SA1016: os.Kill cannot be trapped (did you mean syscall.SIGTERM?) (staticcheck)
signal.Notify(c, os.Interrupt, os.Kill)
Previously, errors from askpass and credential storage were being
ignored, causing git clone/fetch to later error with hard-to-read
errors.
Now the error indicates the credential refresh as the problem, and
does not try to sync.
The "origin" remote is implicitly used as the basis for relative-paths
in submodules. It's very subtly documented, and I have no idea if there
are other places where it is used. It seems git really expects it to
exist, so let's just do that.
This ensures we do not miss events. E.g.
before:
t0: hash changes to X
t1: send webhook(X), waiting for response
t2: hash changes to Y
t3: queue next webhook(Y) but can't send because previous is not done
t4: crash
t5: restart
t6: find repo at hash Y
no webhook(Y) was sent.
after:
t0: hash changes to X
t1: send webhook(X), waiting for response
t2: hash changes to Y
t3: queue next webhook(Y) but can't send because previous is not done
t4: crash
t5: restart
t6: find repo at hash Y
t7: send webhook(Y), waiting for response
AlanZhang2002
Tim Hockin
Liam Wyllie
Sam Dowell
Yves Galante
Aleksandar Markovski
Jiwon Park
justinsb
Kubernetes Prow Robot
Karl Isenberg