abra/cli/app/errors.go

143 lines
3.8 KiB
Go

package app
import (
"context"
"strconv"
"strings"
"time"
"coopcloud.tech/abra/cli/internal"
"coopcloud.tech/abra/pkg/autocomplete"
"coopcloud.tech/abra/pkg/client"
"coopcloud.tech/abra/pkg/config"
"coopcloud.tech/abra/pkg/recipe"
stack "coopcloud.tech/abra/pkg/upstream/stack"
"github.com/docker/docker/api/types"
"github.com/docker/docker/api/types/filters"
dockerClient "github.com/docker/docker/client"
"github.com/sirupsen/logrus"
"github.com/urfave/cli"
)
var appErrorsCommand = cli.Command{
Name: "errors",
Usage: "List errors for a deployed app",
Description: `
This command lists errors for a deployed app.
This is a best-effort implementation and an attempt to gather a number of tips
& tricks for finding errors together into one convenient command. When an app
is failing to deploy or having issues, it could be a lot of things.
This command currently takes into account:
Is the service deployed?
Is the service killed by an OOM error?
Is the service reporting an error (like in "ps --no-trunc" output)
Is the service healthcheck failing? what are the healthcheck logs?
Got any more ideas? Please let us know:
https://git.coopcloud.tech/coop-cloud/organising/issues/new/choose
This command is best accompanied by "abra app logs <app>" which may reveal
further information which can help you debug the cause of an app failure via
the logs.
`,
Aliases: []string{"e"},
Flags: []cli.Flag{
internal.DebugFlag,
internal.NoInputFlag,
internal.WatchFlag,
},
Before: internal.SubCommandBefore,
BashComplete: autocomplete.AppNameComplete,
Action: func(c *cli.Context) error {
app := internal.ValidateApp(c)
cl, err := client.New(app.Server)
if err != nil {
logrus.Fatal(err)
}
isDeployed, _, err := stack.IsDeployed(context.Background(), cl, app.StackName())
if err != nil {
logrus.Fatal(err)
}
if !isDeployed {
logrus.Fatalf("%s is not deployed?", app.Name)
}
if !internal.Watch {
if err := checkErrors(c, cl, app); err != nil {
logrus.Fatal(err)
}
return nil
}
for {
if err := checkErrors(c, cl, app); err != nil {
logrus.Fatal(err)
}
time.Sleep(2 * time.Second)
}
return nil
},
}
func checkErrors(c *cli.Context, cl *dockerClient.Client, app config.App) error {
recipe, err := recipe.Get(app.Type)
if err != nil {
return err
}
for _, service := range recipe.Config.Services {
filters := filters.NewArgs()
filters.Add("name", service.Name)
containers, err := cl.ContainerList(context.Background(), types.ContainerListOptions{Filters: filters})
if err != nil {
return err
}
if len(containers) == 0 {
logrus.Warnf("%s is not up, something seems wrong", service.Name)
continue
}
container := containers[0]
containerState, err := cl.ContainerInspect(context.Background(), container.ID)
if err != nil {
logrus.Fatal(err)
}
if containerState.State.OOMKilled {
logrus.Warnf("%s has been killed due to an out of memory error", service.Name)
}
if containerState.State.Error != "" {
logrus.Warnf("%s reports this error: %s", service.Name, containerState.State.Error)
}
if containerState.State.Health != nil {
if containerState.State.Health.Status != "healthy" {
logrus.Warnf("%s healthcheck status is %s", service.Name, containerState.State.Health.Status)
logrus.Warnf("%s healthcheck has failed %s times", service.Name, strconv.Itoa(containerState.State.Health.FailingStreak))
for _, log := range containerState.State.Health.Log {
logrus.Warnf("%s healthcheck logs: %s", service.Name, strings.TrimSpace(log.Output))
}
}
}
}
return nil
}
func getServiceName(names []string) string {
containerName := strings.Join(names, " ")
trimmed := strings.TrimPrefix(containerName, "/")
return strings.Split(trimmed, ".")[0]
}