- Notifications
You must be signed in to change notification settings - Fork928
feat: support adjusting child proc oom scores#12655
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to ourterms of service andprivacy statement. We’ll occasionally send you account related emails.
Already on GitHub?Sign in to your account
Merged
Uh oh!
There was an error while loading.Please reload this page.
Merged
Changes fromall commits
Commits
Show all changes
10 commits Select commitHold shift + click to select a range
02eb747 fix: adjust process oom_score_adj to 0
sreyab3a787a fmt
sreyaf37242b oom score
sreya403a637 fmt
sreyaa214974 lint
sreya54b7f4c Merge branch 'main' into jon/procprio
sreya17714ed debounce logs
sreya561279b lint
sreya2ddeb8e fmt
sreya9f5f63c pass through env var
sreyaFile filter
Filter by extension
Conversations
Failed to load comments.
Loading
Uh oh!
There was an error while loading.Please reload this page.
Jump to
Jump to file
Failed to load files.
Loading
Uh oh!
There was an error while loading.Please reload this page.
Diff view
Diff view
There are no files selected for viewing
173 changes: 148 additions & 25 deletionsagent/agent.go
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -62,7 +62,10 @@ const ( | ||
| // EnvProcPrioMgmt determines whether we attempt to manage | ||
| // process CPU and OOM Killer priority. | ||
| const ( | ||
| EnvProcPrioMgmt = "CODER_PROC_PRIO_MGMT" | ||
| EnvProcOOMScore = "CODER_PROC_OOM_SCORE" | ||
| ) | ||
| type Options struct { | ||
| Filesystem afero.Fs | ||
| @@ -1575,10 +1578,31 @@ func (a *agent) manageProcessPriorityUntilGracefulShutdown() { | ||
| a.processManagementTick = ticker.C | ||
| } | ||
| oomScore := unsetOOMScore | ||
| if scoreStr, ok := a.environmentVariables[EnvProcOOMScore]; ok { | ||
| score, err := strconv.Atoi(strings.TrimSpace(scoreStr)) | ||
| if err == nil && score >= -1000 && score <= 1000 { | ||
| oomScore = score | ||
sreya marked this conversation as resolved. Show resolvedHide resolvedUh oh!There was an error while loading.Please reload this page. | ||
| } else { | ||
| a.logger.Error(ctx, "invalid oom score", | ||
| slog.F("min_value", -1000), | ||
| slog.F("max_value", 1000), | ||
| slog.F("value", scoreStr), | ||
| ) | ||
| } | ||
| } | ||
| debouncer := &logDebouncer{ | ||
| logger: a.logger, | ||
| messages: map[string]time.Time{}, | ||
| interval: time.Minute, | ||
| } | ||
| for { | ||
| procs, err := a.manageProcessPriority(ctx, debouncer, oomScore) | ||
| // Avoid spamming the logs too often. | ||
| if err != nil { | ||
| debouncer.Error(ctx, "manage process priority", | ||
| slog.Error(err), | ||
| ) | ||
| } | ||
| @@ -1594,42 +1618,51 @@ func (a *agent) manageProcessPriorityUntilGracefulShutdown() { | ||
| } | ||
| } | ||
| // unsetOOMScore is set to an invalid OOM score to imply an unset value. | ||
| const unsetOOMScore = 1001 | ||
| func (a *agent) manageProcessPriority(ctx context.Context, debouncer *logDebouncer, oomScore int) ([]*agentproc.Process, error) { | ||
| const ( | ||
| niceness = 10 | ||
| ) | ||
| // We fetch the agent score each time because it's possible someone updates the | ||
| // value after it is started. | ||
| agentScore, err := a.getAgentOOMScore() | ||
| if err != nil { | ||
| agentScore = unsetOOMScore | ||
| } | ||
| if oomScore == unsetOOMScore && agentScore != unsetOOMScore { | ||
| // If the child score has not been explicitly specified we should | ||
| // set it to a score relative to the agent score. | ||
| oomScore = childOOMScore(agentScore) | ||
| } | ||
| procs, err := agentproc.List(a.filesystem, a.syscaller) | ||
| if err != nil { | ||
| return nil, xerrors.Errorf("list: %w", err) | ||
| } | ||
| modProcs := []*agentproc.Process{} | ||
| for _, proc := range procs { | ||
| containsFn := func(e string) bool { | ||
| contains := strings.Contains(proc.Cmd(), e) | ||
| return contains | ||
| } | ||
| // If the process is prioritized we should adjust | ||
| // it's oom_score_adj and avoid lowering its niceness. | ||
| if slices.ContainsFunc(prioritizedProcs, containsFn) { | ||
| continue | ||
| } | ||
| score, niceErr := proc.Niceness(a.syscaller) | ||
| if niceErr != nil && !xerrors.Is(niceErr, os.ErrPermission) { | ||
| debouncer.Warn(ctx, "unable to get proc niceness", | ||
| slog.F("cmd", proc.Cmd()), | ||
| slog.F("pid", proc.PID), | ||
| slog.Error(niceErr), | ||
| ) | ||
| continue | ||
| } | ||
| @@ -1643,15 +1676,31 @@ func (a *agent) manageProcessPriority(ctx context.Context) ([]*agentproc.Process | ||
| continue | ||
| } | ||
| if niceErr == nil { | ||
| err := proc.SetNiceness(a.syscaller, niceness) | ||
| if err != nil && !xerrors.Is(err, os.ErrPermission) { | ||
| debouncer.Warn(ctx, "unable to set proc niceness", | ||
| slog.F("cmd", proc.Cmd()), | ||
| slog.F("pid", proc.PID), | ||
| slog.F("niceness", niceness), | ||
| slog.Error(err), | ||
| ) | ||
| } | ||
| } | ||
| // If the oom score is valid and it's not already set and isn't a custom value set by another process then it's ok to update it. | ||
| if oomScore != unsetOOMScore && oomScore != proc.OOMScoreAdj && !isCustomOOMScore(agentScore, proc) { | ||
| oomScoreStr := strconv.Itoa(oomScore) | ||
| err := afero.WriteFile(a.filesystem, fmt.Sprintf("/proc/%d/oom_score_adj", proc.PID), []byte(oomScoreStr), 0o644) | ||
| if err != nil && !xerrors.Is(err, os.ErrPermission) { | ||
| debouncer.Warn(ctx, "unable to set oom_score_adj", | ||
| slog.F("cmd", proc.Cmd()), | ||
| slog.F("pid", proc.PID), | ||
| slog.F("score", oomScoreStr), | ||
| slog.Error(err), | ||
| ) | ||
| } | ||
| } | ||
| modProcs = append(modProcs, proc) | ||
| } | ||
| return modProcs, nil | ||
| @@ -2005,3 +2054,77 @@ func PrometheusMetricsHandler(prometheusRegistry *prometheus.Registry, logger sl | ||
| } | ||
| }) | ||
| } | ||
| // childOOMScore returns the oom_score_adj for a child process. It is based | ||
| // on the oom_score_adj of the agent process. | ||
| func childOOMScore(agentScore int) int { | ||
| // If the agent has a negative oom_score_adj, we set the child to 0 | ||
| // so it's treated like every other process. | ||
| if agentScore < 0 { | ||
| return 0 | ||
| } | ||
| // If the agent is already almost at the maximum then set it to the max. | ||
| if agentScore >= 998 { | ||
| return 1000 | ||
| } | ||
| // If the agent oom_score_adj is >=0, we set the child to slightly | ||
| // less than the maximum. If users want a different score they set it | ||
| // directly. | ||
| return 998 | ||
| } | ||
| func (a *agent) getAgentOOMScore() (int, error) { | ||
| scoreStr, err := afero.ReadFile(a.filesystem, "/proc/self/oom_score_adj") | ||
| if err != nil { | ||
| return 0, xerrors.Errorf("read file: %w", err) | ||
| } | ||
| score, err := strconv.Atoi(strings.TrimSpace(string(scoreStr))) | ||
| if err != nil { | ||
| return 0, xerrors.Errorf("parse int: %w", err) | ||
| } | ||
| return score, nil | ||
| } | ||
| // isCustomOOMScore checks to see if the oom_score_adj is not a value that would | ||
| // originate from an agent-spawned process. | ||
| func isCustomOOMScore(agentScore int, process *agentproc.Process) bool { | ||
| score := process.OOMScoreAdj | ||
| return agentScore != score && score != 1000 && score != 0 && score != 998 | ||
| } | ||
| // logDebouncer skips writing a log for a particular message if | ||
| // it's been emitted within the given interval duration. | ||
| // It's a shoddy implementation used in one spot that should be replaced at | ||
| // some point. | ||
| type logDebouncer struct { | ||
| logger slog.Logger | ||
| messages map[string]time.Time | ||
| interval time.Duration | ||
| } | ||
| func (l *logDebouncer) Warn(ctx context.Context, msg string, fields ...any) { | ||
| l.log(ctx, slog.LevelWarn, msg, fields...) | ||
| } | ||
| func (l *logDebouncer) Error(ctx context.Context, msg string, fields ...any) { | ||
| l.log(ctx, slog.LevelError, msg, fields...) | ||
| } | ||
| func (l *logDebouncer) log(ctx context.Context, level slog.Level, msg string, fields ...any) { | ||
| // This (bad) implementation assumes you wouldn't reuse the same msg | ||
| // for different levels. | ||
| if last, ok := l.messages[msg]; ok && time.Since(last) < l.interval { | ||
| return | ||
| } | ||
| switch level { | ||
| case slog.LevelWarn: | ||
| l.logger.Warn(ctx, msg, fields...) | ||
| case slog.LevelError: | ||
| l.logger.Error(ctx, msg, fields...) | ||
| } | ||
| l.messages[msg] = time.Now() | ||
| } | ||
84 changes: 78 additions & 6 deletionsagent/agent_test.go
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.Learn more about bidirectional Unicode characters
10 changes: 8 additions & 2 deletionsagent/agentproc/agentproctest/proc.go
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.Learn more about bidirectional Unicode characters
Oops, something went wrong.
Uh oh!
There was an error while loading.Please reload this page.
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.