CLI: an option to remove zero size objects

* as in: 'ais space-cleanup --rm-zero-size' * separately, gitlab short 32m Signed-off-by: Alex Aizman <[email protected]>
NVIDIA · Nov 26, 2024 · d63e1c2 · d63e1c2
1 parent 44f77df
commit d63e1c2
Show file tree

Hide file tree

Showing 11 changed files with 49 additions and 28 deletions.
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -57,7 +57,7 @@ include:
   stage: test-short
   tags:
     - $RUNNER_TAG
-  timeout: 30m
+  timeout: 32m
   <<: *default_only_def
   except:
     variables:
@@ -69,7 +69,7 @@ include:
   stage: test-short
   tags:
     - $RUNNER_TAG
-  timeout: 30m
+  timeout: 32m
   rules:
     - if: '$CI_PIPELINE_SOURCE == "merge_request_event" || $CI_COMMIT_BRANCH == "main"'
       when: manual
@@ -81,7 +81,7 @@ include:
   stage: test-short
   tags:
     - $RUNNER_TAG
-  timeout: 30m
+  timeout: 32m
   rules:
     - if: '$CI_PIPELINE_SOURCE == "schedule" || $CI_PIPELINE_SOURCE == "web"'
     - if: '$CI_PIPELINE_SOURCE == "merge_request_event" || $CI_COMMIT_BRANCH == "main"'
@@ -500,12 +500,12 @@ test:short:python-etl:
   script:
     - cd python
     - make python_etl_tests
-    
+
 
 #  e.g. RE: "ETLBucket|ETLConnectionError|ETLInitCode" (or any other regex to select tests)
 test:short:assorted:k8s:
   extends: .test_k8s_short_template
-  timeout: 30m
+  timeout: 32m
   rules:
     - if: '$CI_PIPELINE_SOURCE == "merge_request_event" || $CI_COMMIT_BRANCH == "main"'
       when: manual

diff --git a/ais/tgtspace.go b/ais/tgtspace.go
@@ -149,12 +149,11 @@ func (t *target) runSpaceCleanup(xargs *xact.ArgsMsg, wg *sync.WaitGroup) fs.Cap
 		t.bcastAsyncIC(msg)
 	}
 	ini := space.IniCln{
+		StatsT:  t.statsT,
 		Xaction: xcln.(*space.XactCln),
 		Config:  cmn.GCO.Get(),
-		StatsT:  t.statsT,
-		Buckets: xargs.Buckets,
 		WG:      wg,
-		Force:   xargs.Force,
+		Args:    xargs,
 	}
 	xcln.AddNotif(&xact.NotifXact{
 		Base: nl.Base{When: core.UponTerm, Dsts: []string{equalIC}, F: t.notifyTerm},

diff --git a/ais/tgtxact.go b/ais/tgtxact.go
@@ -135,7 +135,7 @@ func (t *target) httpxput(w http.ResponseWriter, r *http.Request) {
 			}
 			return
 		}
-		// all other "startables"
+		// all other _startable_ xactions
 		xid, err := t.xstart(&xargs, bck, msg)
 		if err != nil {
 			t.writeErr(w, r, err)

diff --git a/cmd/cli/cli/const.go b/cmd/cli/cli/const.go
@@ -40,7 +40,7 @@ const (
 	commandShow = "show"
 )
 
-// advanced command and subcommands
+// 'ais advanced' subcommands
 const (
 	cmdGenShards     = "gen-shards"
 	cmdPreload       = "preload"
@@ -50,6 +50,8 @@ const (
 	cmdRotateLogs    = "rotate-logs"
 )
 
+const advancedUsageOnly = "(caution: advanced usage only)"
+
 // - 2nd level subcommands (mostly, verbs)
 // - show subcommands (`show <what>`)
 // - 3rd level subcommands
@@ -502,6 +504,8 @@ var (
 			indent1 + "\t(tip: check 'ais config cluster lru.dont_evict_time' as well)",
 	}
 
+	rmZeroSizeFlag = cli.BoolFlag{Name: "rm-zero-size", Usage: "remove zero size objects " + advancedUsageOnly}
+
 	// units enum { unitsIEC, unitsSI, unitsRaw }
 	unitsFlag = cli.StringFlag{
 		Name: "units",
@@ -1059,11 +1063,11 @@ var (
 	}
 	nonElectableFlag = cli.BoolFlag{
 		Name:  "non-electable",
-		Usage: "this proxy must not be elected as primary (advanced use)",
+		Usage: "this proxy must not be elected as primary " + advancedUsageOnly,
 	}
 	noRebalanceFlag = cli.BoolFlag{
 		Name:  "no-rebalance",
-		Usage: "do _not_ run global rebalance after putting node in maintenance (caution: advanced usage only!)",
+		Usage: "do _not_ run global rebalance after putting node in maintenance " + advancedUsageOnly,
 	}
 	mountpathLabelFlag = cli.StringFlag{
 		Name: "label",

diff --git a/cmd/cli/cli/storage_hdlr.go b/cmd/cli/cli/storage_hdlr.go
@@ -92,7 +92,7 @@ var (
 			{
 				Name: cmdMpathRescanDisks,
 				Usage: "re-resolve (mountpath, filesystem) to its underlying disk(s) and revalidate the disks\n" +
-					indent1 + "\t(advanced use only)",
+					indent1 + "\t" + advancedUsageOnly,
 				ArgsUsage:    nodeMountpathPairArgument,
 				Flags:        mpathCmdsFlags["default"],
 				Action:       mpathRescanHandler,
@@ -112,12 +112,13 @@ var (
 var (
 	cleanupFlags = []cli.Flag{
 		forceClnFlag,
+		rmZeroSizeFlag,
 		waitFlag,
 		waitJobXactFinishedFlag,
 	}
 	cleanupCmd = cli.Command{
 		Name:         cmdStgCleanup,
-		Usage:        "perform storage cleanup: remove deleted objects and old/obsolete workfiles",
+		Usage:        "remove deleted objects and old/obsolete workfiles; remove misplaced objects; optionally, remove zero size objects",
 		ArgsUsage:    listAnyCommandArgument,
 		Flags:        cleanupFlags,
 		Action:       cleanupStorageHandler,
@@ -229,8 +230,14 @@ func cleanupStorageHandler(c *cli.Context) (err error) {
 		}
 	}
 
-	force := flagIsSet(c, forceFlag)
+	// xargs
+	force := flagIsSet(c, forceClnFlag)
 	xargs := xact.ArgsMsg{Kind: apc.ActStoreCleanup, Bck: bck, Force: force}
+	if flagIsSet(c, rmZeroSizeFlag) {
+		xargs.Flags = xact.XrmZeroSize
+	}
+
+	// do
 	if id, err = api.StartXaction(apiBP, &xargs, ""); err != nil {
 		return
 	}

diff --git a/cmd/cli/go.mod b/cmd/cli/go.mod
@@ -3,7 +3,7 @@ module github.com/NVIDIA/aistore/cmd/cli
 go 1.23.2
 
 require (
-	github.com/NVIDIA/aistore v1.3.26-0.20241119190643-7c6f3181bf9f
+	github.com/NVIDIA/aistore v1.3.26-0.20241126181423-44f77dfe5637
 	github.com/fatih/color v1.18.0
 	github.com/json-iterator/go v1.1.12
 	github.com/onsi/ginkgo/v2 v2.21.0

diff --git a/cmd/cli/go.sum b/cmd/cli/go.sum
@@ -1,7 +1,7 @@
 code.cloudfoundry.org/bytefmt v0.0.0-20190710193110-1eb035ffe2b6/go.mod h1:wN/zk7mhREp/oviagqUXY3EwuHhWyOvAdsn5Y4CzOrc=
 github.com/BurntSushi/toml v1.4.0/go.mod h1:ukJfTF/6rtPPRCnwkur4qwRxa8vTRFBF0uk2lLoLwho=
-github.com/NVIDIA/aistore v1.3.26-0.20241119190643-7c6f3181bf9f h1:+Cl8s4DwoXEX9XKqXZBvDt+Fhbv4zGcu8TFm9j9vujc=
-github.com/NVIDIA/aistore v1.3.26-0.20241119190643-7c6f3181bf9f/go.mod h1:I3l9tS47EY4ODo7mQQGsFFrIjHa6mwnNRF4t1LXKCvA=
+github.com/NVIDIA/aistore v1.3.26-0.20241126181423-44f77dfe5637 h1:mbCYjPrmGzEKSIWP3aS9mY2DhKb77Wbxmwk3ybM1TP0=
+github.com/NVIDIA/aistore v1.3.26-0.20241126181423-44f77dfe5637/go.mod h1:/6lHPQptxjseMk9B2Zv3wcFkXsyZdr4YCsbfjzHe7r8=
 github.com/OneOfOne/xxhash v1.2.8 h1:31czK/TI9sNkxIKfaUfGlU47BAxQ0ztGgd9vPyqimf8=
 github.com/OneOfOne/xxhash v1.2.8/go.mod h1:eZbhyaAYD41SGSSsnmcpxVoRiQ/MPUTjUdIIOT9Um7Q=
 github.com/VividCortex/ewma v1.1.1/go.mod h1:2Tkkvm3sRDVXaiyucHiACn4cqf7DpdyLvmxzcbUokwA=

diff --git a/docs/cli/storage.md b/docs/cli/storage.md
@@ -58,14 +58,15 @@ $ find . -type f -name "*.md" | xargs grep "ais.*mountpath"
 ```console
 $ ais storage cleanup --help
 NAME:
-   ais storage cleanup - perform storage cleanup: remove deleted objects and old/obsolete workfiles
+   ais storage cleanup - remove deleted objects and old/obsolete workfiles; remove misplaced objects; optionally, remove zero size objects
 
 USAGE:
    ais storage cleanup [command options] PROVIDER:[//BUCKET_NAME]
 
 OPTIONS:
    --force, -f      disregard interrupted rebalance and possibly other conditions preventing full cleanup
                     (tip: check 'ais config cluster lru.dont_evict_time' as well)
+   --rm-zero-size   remove zero-size objects (caution: advanced usage only)
    --wait           wait for an asynchronous operation to finish (optionally, use '--timeout' to limit the waiting time)
    --timeout value  maximum time to wait for a job to finish; if omitted: wait forever or until Ctrl-C;
                     valid time units: ns, us (or µs), ms, s (default), m, h

diff --git a/space/cleanup.go b/space/cleanup.go
@@ -32,12 +32,11 @@ import (
 
 type (
 	IniCln struct {
+		StatsT  stats.Tracker
 		Config  *cmn.Config
 		Xaction *XactCln
-		StatsT  stats.Tracker
-		Buckets []cmn.Bck // optional list of specific buckets to cleanup
 		WG      *sync.WaitGroup
-		Force   bool
+		Args    *xact.ArgsMsg
 	}
 	XactCln struct {
 		xact.Base
@@ -209,10 +208,10 @@ func (p *clnP) rmMisplaced() bool {
 		why = g.Xact.String() + " is running"
 	case g.Interrupted:
 		why = "rebalance interrupted"
-		ok = p.ini.Force
+		ok = p.ini.Args.Force
 	case g.Restarted:
 		why = "node restarted"
-		ok = p.ini.Force
+		ok = p.ini.Args.Force
 	case l.Xact != nil:
 		why = l.Xact.String() + " is running"
 	case l.Interrupted:
@@ -241,7 +240,7 @@ func (j *clnJ) String() string {
 	sb.WriteString(j.ini.Xaction.String())
 	sb.WriteString(": jog-")
 	sb.WriteString(j.mi.String())
-	if j.ini.Force {
+	if j.ini.Args.Force {
 		sb.WriteString("-with-force")
 	}
 	return sb.String()
@@ -262,8 +261,8 @@ func (j *clnJ) run(providers []string) {
 	}
 
 	// traverse
-	if len(j.ini.Buckets) != 0 {
-		size, err = j.jogBcks(j.ini.Buckets)
+	if len(j.ini.Args.Buckets) != 0 {
+		size, err = j.jogBcks(j.ini.Args.Buckets)
 	} else {
 		size, err = j.jog(providers)
 	}
@@ -489,6 +488,15 @@ func (j *clnJ) visitObj(fqn string, lom *core.LOM) {
 		if lom.HasCopies() {
 			j.rmExtraCopies(lom)
 		}
+		if lom.Lsize() == 0 {
+			if j.ini.Args.Flags&xact.XrmZeroSize == xact.XrmZeroSize {
+				if ecode, err := core.T.DeleteObject(lom, false /*evict*/); err != nil {
+					nlog.Errorln("failed to remove zero-size", lom.Cname(), "err:", err, "code:", ecode)
+				} else {
+					nlog.Warningln("removed zero-size", lom.Cname())
+				}
+			}
+		}
 		return
 	}
 	if lom.IsCopy() {

diff --git a/space/space_test.go b/space/space_test.go
@@ -22,6 +22,7 @@ import (
 	"github.com/NVIDIA/aistore/hk"
 	"github.com/NVIDIA/aistore/space"
 	"github.com/NVIDIA/aistore/tools/trand"
+	"github.com/NVIDIA/aistore/xact"
 	"github.com/NVIDIA/aistore/xact/xreg"
 	. "github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
@@ -360,6 +361,7 @@ func newInitStoreCln() *space.IniCln {
 		Xaction: xcln,
 		Config:  cmn.GCO.Get(),
 		StatsT:  mock.NewStatsTracker(),
+		Args:    &xact.ArgsMsg{},
 	}
 }
 

diff --git a/xact/api.go b/xact/api.go
@@ -51,7 +51,7 @@ const (
 
 // ArgsMsg.Flags
 const (
-	XrmZeroSize = 1 << iota
+	XrmZeroSize = 1 << iota // usage: x-cleanup (apc.ActStoreCleanup) to remove zero size objects
 )
 
 type (
-Original file line number
+Diff line change
@@ Expand Up / @@ -51,7 +51,7 @@ const ( @@
     // ArgsMsg.Flags
     const (
-    	XrmZeroSize = 1 << iota
+    	XrmZeroSize = 1 << iota // usage: x-cleanup (apc.ActStoreCleanup) to remove zero size objects
     )
     type (
@@ Expand Down @@