What causes "failed to get mounts of %q" in dagger/dagger?

Retry the Merge operation — prepareKey is a fresh identity.NewID() each time, so a retry allocates a new snapshot

failed to get mounts of %q

Error message

failed to get mounts of %q

What it means

Merge wraps failures from sn.Mounts(ctx, prepareKey) with "failed to get mounts of %q". After successfully preparing an intermediate snapshot, the code asks the snapshotter for its mount configuration to apply the diffs onto; if the snapshotter cannot compute mounts (snapshot vanished, backend error), this error is returned. The %q is the destination key, not the internal prepareKey.

Source

Thrown at engine/snapshots/merge.go:122

		}
		ctx = leaseCtx
	}
	if leaseID, ok := leases.FromContext(ctx); !ok || leaseID == "" {
		leaseCtx, done, err := WithLease(ctx, sn.lm, MakeTemporary)
		if err != nil {
			return errors.Wrap(err, "failed to create temporary lease for view mounts during merge")
		}
		defer done(context.TODO())
		ctx = leaseCtx
	}

	prepareKey := identity.NewID()
	if err := sn.Prepare(ctx, prepareKey, baseKey); err != nil {
		return errors.Wrapf(err, "failed to prepare %q", key)
	}
	applyMounts, err := sn.Mounts(ctx, prepareKey)
	if err != nil {
		return errors.Wrapf(err, "failed to get mounts of %q", key)
	}

	usage, err := sn.diffApply(ctx, applyMounts, diffs...)
	if err != nil {
		return errors.Wrap(err, "failed to apply diffs")
	}
	if err := sn.Commit(ctx, key, prepareKey, withMergeUsage(usage)); err != nil {
		return errors.Wrapf(err, "failed to commit %q", key)
	}
	return nil
}

func (sn *mergeSnapshotter) Usage(ctx context.Context, key string) (snapshots.Usage, error) {
	if info, err := sn.Stat(ctx, key); err != nil {
		return snapshots.Usage{}, err
	} else if usage, ok, err := mergeUsageOf(info); err != nil {
		return snapshots.Usage{}, err
	} else if ok {

View on GitHub (pinned to 82ba2681db)

Solutions

Retry the Merge operation — prepareKey is a fresh identity.NewID() each time, so a retry allocates a new snapshot
Check containerd daemon logs for snapshotter/Mounts errors around the timestamp
Verify no aggressive GC/pruning job runs concurrently with merges
Confirm the snapshotter driver (overlayfs/stargz/native) is healthy on the node

Defensive patterns

Strategy: retry

Validate before calling

// prepareKey is internal; validate the diff endpoints exist
for _, d := range diffs {
    if _, err := sn.Stat(ctx, d.Lower); err != nil { return err }
}

Try / catch

if err := sn.Merge(ctx, key, diffs); err != nil {
    if strings.Contains(err.Error(), "failed to get mounts of") {
        return retryMerge(ctx, key, diffs) // fresh prepareKey on retry
    }
    return err
}

Prevention

Retry Merge on transient mount errors — a new snapshot is prepared each attempt
Disable concurrent GC pruning while merges run
Watch containerd logs for snapshotter mount failures
Keep the snapshotter driver healthy (kernel module, backing fs)

When it happens

Trigger: Merge(ctx, key, diffs) where the snapshotter's Mounts call on the freshly prepared prepareKey fails — e.g. snapshot was deleted between Prepare and Mounts, or the overlay backend cannot construct the mount options.

Common situations: Concurrent garbage collection removing prepareKey; containerd snapshotter wedged after node restart; overlayfs kernel/module issues on the host.

Related errors

AI-assisted analysis of dagger/dagger@82ba2681db (2026-09-05). Data as JSON: /api/errors/07ca0ce022b3ede9. Report an issue: GitHub.