Imporve healing and reporting (#11312)

* Provide information on *actively* healing, buckets healed/queued, objects healed/failed.
* Add concurrent healing of multiple sets (typically on startup).
* Add bucket level resume, so restarts will only heal non-healed buckets.
* Print summary after healing a disk is done.
This commit is contained in:
Klaus Post
2021-03-04 14:36:23 -08:00
committed by GitHub
parent 97e7a902d0
commit fa9cf1251b
36 changed files with 1357 additions and 307 deletions
+15 -7
View File
@@ -239,13 +239,6 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
storageDisks := er.getDisks()
storageEndpoints := er.getEndpoints()
// List of disks having latest version of the object er.meta
// (by modtime).
latestDisks, modTime := listOnlineDisks(storageDisks, partsMetadata, errs)
// List of disks having all parts as per latest er.meta.
availableDisks, dataErrs := disksWithAllParts(ctx, latestDisks, partsMetadata, errs, bucket, object, scanMode)
// Initialize heal result object
result = madmin.HealResultItem{
Type: madmin.HealItemObject,
@@ -256,6 +249,21 @@ func (er erasureObjects) healObject(ctx context.Context, bucket string, object s
DataBlocks: len(storageDisks) - er.defaultParityCount,
}
if !opts.NoLock {
lk := er.NewNSLock(bucket, object)
if ctx, err = lk.GetLock(ctx, globalOperationTimeout); err != nil {
return result, err
}
defer lk.Unlock()
}
// List of disks having latest version of the object er.meta
// (by modtime).
latestDisks, modTime := listOnlineDisks(storageDisks, partsMetadata, errs)
// List of disks having all parts as per latest er.meta.
availableDisks, dataErrs := disksWithAllParts(ctx, latestDisks, partsMetadata, errs, bucket, object, scanMode)
// Loop to find number of disks with valid data, per-drive
// data state and a list of outdated disks on which data needs
// to be healed.