Merge remote-tracking branch 'origin/master' into ckochhof/dev/master…

…/daos-15596

Required-githooks: true

.github/workflows/trivy.yml

@@ -1,6 +1,9 @@
 name: Trivy scan
 
 on:
+  workflow_dispatch:
+  push:
+    branches: ["master", "release/**"]
   pull_request:
     branches: ["master", "release/**"]
 

docs/admin/env_variables.md

@@ -44,6 +44,7 @@ Environment variables in this section only apply to the server side.
 |DAOS\_MD\_CAP         |Size of a metadata pmem pool/file in MBs. INTEGER. Default to 128 MB.|
 |DAOS\_START\_POOL\_SVC|Determines whether to start existing pool services when starting a daos\_server. BOOL. Default to true.|
 |CRT\_DISABLE\_MEM\_PIN|Disable memory pinning workaround on a server side. BOOL. Default to 0.|
+|CRT\_EVENT\_DELAY|Delay in seconds before handling each CaRT event. INTEGER. Default to 10 s. A longer delay enables batching of successive CaRT events, leading to fewer pool map changes when multiple engines become unavailable at around the same time.|
 |DAOS\_SCHED\_PRIO\_DISABLED|Disable server ULT prioritizing. BOOL. Default to 0.|
 |DAOS\_SCHED\_RELAX\_MODE|The mode of CPU relaxing on idle. "disabled":disable relaxing; "net":wait on network request for INTVL; "sleep":sleep for INTVL. STRING. Default to "net"|
 |DAOS\_SCHED\_RELAX\_INTVL|CPU relax interval in milliseconds. INTEGER. Default to 1 ms.|

src/client/dfuse/pil4dfs/int_dfs.c

@@ -6302,6 +6302,14 @@ new_mmap(void *addr, size_t length, int prot, int flags, int fd, off_t offset)
 
 	atomic_fetch_add_relaxed(&num_mmap, 1);
 
+	if ((fd < FD_FILE_BASE) && (fd_directed >= FD_FILE_BASE) && d_compatible_mode) {
+		/* DAOS-14494: Force the kernel to update the size before mapping. */
+		rc = next_fxstat(1, fd, &stat_buf);
+		if (rc == -1)
+			return MAP_FAILED;
+		return next_mmap(addr, length, prot, flags, fd, offset);
+	}
+
 	addr_ret = next_mmap(addr, length, prot, flags | MAP_ANONYMOUS, -1, offset);
 	if (addr_ret == MAP_FAILED)
 		return MAP_FAILED;
@@ -7086,10 +7094,8 @@ init_myhook(void)
 
 	register_a_hook("libc", "fcntl", (void *)new_fcntl, (long int *)(&libc_fcntl));
 
-	if (d_compatible_mode == false) {
-		register_a_hook("libc", "mmap", (void *)new_mmap, (long int *)(&next_mmap));
-		register_a_hook("libc", "munmap", (void *)new_munmap, (long int *)(&next_munmap));
-	}
+	register_a_hook("libc", "mmap", (void *)new_mmap, (long int *)(&next_mmap));
+	register_a_hook("libc", "munmap", (void *)new_munmap, (long int *)(&next_munmap));
 
 	register_a_hook("libc", "exit", (void *)new_exit, (long int *)(&next_exit));
 	register_a_hook("libc", "dup3", (void *)new_dup3, (long int *)(&libc_dup3));

src/client/pydaos/pydaos_shim.c

@@ -70,6 +70,7 @@ static daos_handle_t eq;
  * Implementations of baseline shim functions
  */
 
+#if 0
 static void
 child_handler(void)
 {
@@ -84,6 +85,7 @@ child_handler(void)
 	if (rc)
 		DL_ERROR(rc, "Failed to re-create global eq");
 }
+#endif
 
 static PyObject *
 __shim_handle__daos_init(PyObject *self, PyObject *args)
@@ -101,11 +103,14 @@ __shim_handle__daos_init(PyObject *self, PyObject *args)
 		return PyLong_FromLong(rc);
 	}
 
+#if 0
+	/** disabled due to DAOS-16637 */
 	rc = pthread_atfork(NULL, NULL, &child_handler);
 	if (rc) {
 		DL_ERROR(rc, "Failed to set atfork handler");
 		return PyLong_FromLong(rc);
 	}
+#endif
 
 	return PyLong_FromLong(rc);
 }

src/control/cmd/dmg/pretty/storage.go

@@ -11,12 +11,8 @@ import (
 	"io"
 	"strings"
 
-	"github.com/dustin/go-humanize"
-	"github.com/pkg/errors"
-
 	"github.com/daos-stack/daos/src/control/lib/control"
 	"github.com/daos-stack/daos/src/control/lib/txtfmt"
-	"github.com/daos-stack/daos/src/control/server/storage"
 )
 
 // printHostStorageMapVerbose generates a human-readable representation of the supplied
@@ -85,44 +81,6 @@ func PrintHostStorageMap(hsm control.HostStorageMap, out io.Writer, opts ...Prin
 	return nil
 }
 
-// PrintHostStorageUsageMap generates a human-readable representation of the supplied
-// HostStorageMap struct and writes utilization info to the supplied io.Writer.
-func PrintHostStorageUsageMap(hsm control.HostStorageMap, out io.Writer) error {
-	if len(hsm) == 0 {
-		return nil
-	}
-
-	hostsTitle := "Hosts"
-	scmTitle := "SCM-Total"
-	scmFreeTitle := "SCM-Free"
-	scmUsageTitle := "SCM-Used"
-	nvmeTitle := "NVMe-Total"
-	nvmeFreeTitle := "NVMe-Free"
-	nvmeUsageTitle := "NVMe-Used"
-
-	tablePrint := txtfmt.NewTableFormatter(hostsTitle, scmTitle, scmFreeTitle,
-		scmUsageTitle, nvmeTitle, nvmeFreeTitle, nvmeUsageTitle)
-	tablePrint.InitWriter(out)
-	table := []txtfmt.TableRow{}
-
-	for _, key := range hsm.Keys() {
-		hss := hsm[key]
-		hosts := getPrintHosts(hss.HostSet.RangedString())
-		row := txtfmt.TableRow{hostsTitle: hosts}
-		storage := hss.HostStorage
-		row[scmTitle] = humanize.Bytes(storage.ScmNamespaces.Total())
-		row[scmFreeTitle] = humanize.Bytes(storage.ScmNamespaces.Free())
-		row[scmUsageTitle] = storage.ScmNamespaces.PercentUsage()
-		row[nvmeTitle] = humanize.Bytes(storage.NvmeDevices.Total())
-		row[nvmeFreeTitle] = humanize.Bytes(storage.NvmeDevices.Free())
-		row[nvmeUsageTitle] = storage.NvmeDevices.PercentUsage()
-		table = append(table, row)
-	}
-
-	tablePrint.Format(table)
-	return nil
-}
-
 func printStorageFormatMapVerbose(hsm control.HostStorageMap, out io.Writer, opts ...PrintConfigOption) error {
 	for _, key := range hsm.Keys() {
 		hss := hsm[key]
@@ -176,153 +134,3 @@ func PrintStorageFormatMap(hsm control.HostStorageMap, out io.Writer, opts ...Pr
 	tablePrint.Format(table)
 	return nil
 }
-
-// NVMe controller namespace ID (NSID) should only be displayed if >= 1. Zero value should be
-// ignored in display output.
-func printSmdDevice(dev *storage.SmdDevice, iw io.Writer, opts ...PrintConfigOption) error {
-	fc := getPrintConfig(opts...)
-
-	trAddr := fmt.Sprintf("TrAddr:%s", dev.Ctrlr.PciAddr)
-	nsID := fmt.Sprintf("NSID:%d", dev.CtrlrNamespaceID)
-	uid := fmt.Sprintf("UUID:%s", dev.UUID)
-	led := fmt.Sprintf("LED:%s", dev.Ctrlr.LedState)
-
-	if fc.LEDInfoOnly {
-		out := trAddr
-		if dev.CtrlrNamespaceID > 0 {
-			out = fmt.Sprintf("%s %s", out, nsID)
-		}
-		if dev.UUID != "" {
-			out = fmt.Sprintf("%s [%s]", out, uid)
-		}
-
-		_, err := fmt.Fprintf(iw, "%s %s\n", out, led)
-		return err
-	}
-
-	out := fmt.Sprintf("%s [%s", uid, trAddr)
-	if dev.CtrlrNamespaceID > 0 {
-		out = fmt.Sprintf("%s %s", out, nsID)
-	}
-	if _, err := fmt.Fprintf(iw, "%s]\n", out); err != nil {
-		return err
-	}
-
-	var hasSysXS string
-	if dev.HasSysXS {
-		hasSysXS = "SysXS "
-	}
-	if _, err := fmt.Fprintf(txtfmt.NewIndentWriter(iw),
-		"Roles:%s %sTargets:%+v Rank:%d State:%s LED:%s\n", &dev.Roles, hasSysXS,
-		dev.TargetIDs, dev.Rank, dev.Ctrlr.NvmeState, dev.Ctrlr.LedState); err != nil {
-		return err
-	}
-
-	return nil
-}
-
-func printSmdPool(pool *control.SmdPool, out io.Writer, opts ...PrintConfigOption) error {
-	ew := txtfmt.NewErrWriter(out)
-	fmt.Fprintf(ew, "Rank:%d Targets:%+v", pool.Rank, pool.TargetIDs)
-	cfg := getPrintConfig(opts...)
-	if cfg.Verbose {
-		fmt.Fprintf(ew, " Blobs:%+v", pool.Blobs)
-	}
-	fmt.Fprintln(ew)
-	return ew.Err
-}
-
-// PrintSmdInfoMap generates a human-readable representation of the supplied
-// HostStorageMap, with a focus on presenting the per-server metadata (SMD) information.
-func PrintSmdInfoMap(omitDevs, omitPools bool, hsm control.HostStorageMap, out io.Writer, opts ...PrintConfigOption) error {
-	w := txtfmt.NewErrWriter(out)
-
-	for _, key := range hsm.Keys() {
-		hss := hsm[key]
-		hosts := getPrintHosts(hss.HostSet.RangedString(), opts...)
-		lineBreak := strings.Repeat("-", len(hosts))
-		fmt.Fprintf(out, "%s\n%s\n%s\n", lineBreak, hosts, lineBreak)
-
-		iw := txtfmt.NewIndentWriter(out)
-		if hss.HostStorage.SmdInfo == nil {
-			fmt.Fprintln(iw, "No SMD info returned")
-			continue
-		}
-
-		if !omitDevs {
-			if len(hss.HostStorage.SmdInfo.Devices) > 0 {
-				fmt.Fprintln(iw, "Devices")
-
-				for _, device := range hss.HostStorage.SmdInfo.Devices {
-					iw1 := txtfmt.NewIndentWriter(iw)
-					if err := printSmdDevice(device, iw1, opts...); err != nil {
-						return err
-					}
-					if device.Ctrlr.HealthStats == nil {
-						continue
-					}
-					if err := printNvmeHealth(device.Ctrlr.HealthStats,
-						txtfmt.NewIndentWriter(iw1), opts...); err != nil {
-						return err
-					}
-					fmt.Fprintln(out)
-				}
-			} else {
-				fmt.Fprintln(iw, "No devices found")
-			}
-		}
-
-		if !omitPools {
-			if len(hss.HostStorage.SmdInfo.Pools) > 0 {
-				fmt.Fprintln(iw, "Pools")
-
-				for uuid, poolSet := range hss.HostStorage.SmdInfo.Pools {
-					iw1 := txtfmt.NewIndentWriter(iw)
-					fmt.Fprintf(iw1, "UUID:%s\n", uuid)
-					iw2 := txtfmt.NewIndentWriter(iw1)
-					for _, pool := range poolSet {
-						if err := printSmdPool(pool, iw2, opts...); err != nil {
-							return err
-						}
-					}
-					fmt.Fprintln(out)
-				}
-			} else {
-				fmt.Fprintln(iw, "No pools with NVMe found")
-			}
-		}
-	}
-
-	return w.Err
-}
-
-// PrintSmdManageResp generates a human-readable representation of the supplied response.
-func PrintSmdManageResp(op control.SmdManageOpcode, resp *control.SmdResp, out, outErr io.Writer, opts ...PrintConfigOption) error {
-	switch op {
-	case control.SetFaultyOp, control.DevReplaceOp:
-		if resp.ResultCount() > 1 {
-			return errors.Errorf("smd-manage %s: unexpected number of results, "+
-				"want %d got %d", op, 1, resp.ResultCount())
-		}
-
-		hem := resp.GetHostErrors()
-		if len(hem) > 0 {
-			for errStr, hostSet := range hem {
-				fmt.Fprintln(outErr, fmt.Sprintf("%s operation failed on %s: %s",
-					op, hostSet.HostSet, errStr))
-			}
-			return nil
-		}
-
-		return PrintHostStorageSuccesses(fmt.Sprintf("%s operation performed", op),
-			resp.HostStorage, out)
-	case control.LedCheckOp, control.LedBlinkOp, control.LedResetOp:
-		if err := PrintResponseErrors(resp, outErr, opts...); err != nil {
-			return err
-		}
-
-		return PrintSmdInfoMap(false, true, resp.HostStorage, out, opts...)
-	default:
-		return errors.Errorf("unsupported opcode %d", op)
-	}
-}