//
// SPDX-License-Identifier: AGPL-3.0
-package main
+package keepbalance
import (
"bytes"
"context"
"crypto/md5"
+ "errors"
"fmt"
"io"
"io/ioutil"
"syscall"
"time"
+ "git.arvados.org/arvados.git/lib/controller/dblock"
"git.arvados.org/arvados.git/sdk/go/arvados"
+ "git.arvados.org/arvados.git/sdk/go/ctxlog"
"git.arvados.org/arvados.git/sdk/go/keepclient"
"github.com/jmoiron/sqlx"
"github.com/sirupsen/logrus"
// subsequent balance operation.
//
// Run should only be called once on a given Balancer object.
-//
-// Typical usage:
-//
-// runOptions, err = (&Balancer{}).Run(config, runOptions)
-func (bal *Balancer) Run(client *arvados.Client, cluster *arvados.Cluster, runOptions RunOptions) (nextRunOptions RunOptions, err error) {
+func (bal *Balancer) Run(ctx context.Context, client *arvados.Client, cluster *arvados.Cluster, runOptions RunOptions) (nextRunOptions RunOptions, err error) {
nextRunOptions = runOptions
+ ctxlog.FromContext(ctx).Info("acquiring active lock")
+ if !dblock.KeepBalanceActive.Lock(ctx, func(context.Context) (*sqlx.DB, error) { return bal.DB, nil }) {
+ // context canceled
+ return
+ }
+ defer dblock.KeepBalanceActive.Unlock()
+
defer bal.time("sweep", "wall clock time to run one full sweep")()
- ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(cluster.Collections.BalanceTimeout.Duration()))
+ ctx, cancel := context.WithDeadline(ctx, time.Now().Add(cluster.Collections.BalanceTimeout.Duration()))
defer cancel()
var lbFile *os.File
}
}
+ mountProblem := false
+ type deviceMount struct {
+ srv *KeepService
+ mnt *KeepMount
+ }
+ deviceMounted := map[string]deviceMount{} // DeviceID -> mount
+ for _, srv := range bal.KeepServices {
+ for _, mnt := range srv.mounts {
+ if first, dup := deviceMounted[mnt.DeviceID]; dup && first.mnt.UUID != mnt.UUID && mnt.DeviceID != "" {
+ bal.logf("config error: device %s is mounted with multiple volume UUIDs: %s on %s, and %s on %s",
+ mnt.DeviceID,
+ first.mnt.UUID, first.srv,
+ mnt.UUID, srv)
+ mountProblem = true
+ continue
+ }
+ deviceMounted[mnt.DeviceID] = deviceMount{srv, mnt}
+ }
+ }
+ if mountProblem {
+ return errors.New("cannot continue with config errors (see above)")
+ }
+
var checkPage arvados.CollectionList
if err = c.RequestAndDecode(&checkPage, "GET", "arvados/v1/collections", nil, arvados.ResourceListParams{
Limit: new(int),