GEODE-10613: Add configurable export directories for export data (#8042)

Author: JinwooHwang

geode-docs/security/implementing_authorization.html.md.erb

@@ -154,7 +154,7 @@ This table classifies the permissions assigned for `gfsh` operations.
 | execute function                       | Defaults to DATA:WRITE. Override `Function.getRequiredPermissions` to change the permission.   |
 | export cluster-configuration           | CLUSTER:READ                     |
 | export config                          | CLUSTER:READ                     |
-| export data                            | CLUSTER:READ                     |
+| export data                            | DATA:READ:RegionName and CLUSTER:WRITE  |
 | export logs                            | CLUSTER:READ                     |
 | export offline-disk-store              | CLUSTER:READ                     |
 | export stack-traces                    | CLUSTER:READ                     |

geode-docs/tools_modules/gfsh/command-pages/export.html.md.erb

@@ -165,6 +165,22 @@ In this scenario, partitioned region data is exported simultaneously on all host
 | <span class="keyword parmname">&#8209;&#8209;dir</span>   | Directory to which the exported data is to be written. Required if &#8209;&#8209;parallel is true. Cannot be specified at the same time as &#8209;&#8209;file.|
 | <span class="keyword parmname">&#8209;&#8209;parallel</span>   | Export local data on each node to a directory on that machine. Available for partitioned regions only.  |
 
+**Export locations:**
+
+The snapshot is written by the member named in `--member`, on that member's host. A member writes
+exports into its own working directory (and sub-directories of it). To export somewhere else, such
+as a mounted backup location, set the `gemfire.export.data.dirs` system property on the member to
+the additional directories, separated by the platform's path separator:
+
+``` pre
+-Dgemfire.export.data.dirs=/mnt/backup/geode:/var/exports/geode
+```
+
+A path containing a `..` segment is not accepted, and a path that resolves outside the configured
+directories is rejected by the member.
+
+**Required permission:** `DATA:READ` on the exported region, plus `CLUSTER:WRITE`.
+
 **Example Commands:**
 
 ``` pre

geode-gfsh/src/integrationTest/java/org/apache/geode/management/internal/cli/commands/ExportDataIntegrationTest.java

@@ -17,6 +17,7 @@
 package org.apache.geode.management.internal.cli.commands;
 
 import static org.apache.geode.cache.Region.SEPARATOR;
+import static org.apache.geode.management.internal.cli.functions.ExportDataFunction.EXPORT_DATA_DIRS_PROPERTY;
 import static org.assertj.core.api.Assertions.assertThat;
 import static org.junit.Assert.assertFalse;
 
@@ -31,6 +32,7 @@
 import org.junit.ClassRule;
 import org.junit.Rule;
 import org.junit.Test;
+import org.junit.contrib.java.lang.system.RestoreSystemProperties;
 import org.junit.rules.TemporaryFolder;
 
 import org.apache.geode.DataSerializable;
@@ -58,6 +60,9 @@ public class ExportDataIntegrationTest {
   @Rule
   public TemporaryFolder tempDir = new TemporaryFolder();
 
+  @Rule
+  public RestoreSystemProperties restoreSystemProperties = new RestoreSystemProperties();
+
   private Region<String, Object> region;
   private Path snapshotFile;
   private Path snapshotDir;
@@ -87,6 +92,8 @@ public void setup() throws Exception {
     region = server.getCache().getRegion(TEST_REGION_NAME);
     loadRegion("value");
     Path basePath = tempDir.getRoot().toPath();
+    // configure the test's temporary folder as an export destination
+    System.setProperty(EXPORT_DATA_DIRS_PROPERTY, basePath.toString());
     snapshotFile = basePath.resolve(SNAPSHOT_FILE);
     snapshotDir = basePath.resolve(SNAPSHOT_DIR);
   }

geode-gfsh/src/integrationTest/java/org/apache/geode/management/internal/cli/commands/ImportDataIntegrationTest.java

@@ -17,6 +17,7 @@
 package org.apache.geode.management.internal.cli.commands;
 
 import static org.apache.geode.cache.Region.SEPARATOR;
+import static org.apache.geode.management.internal.cli.functions.ExportDataFunction.EXPORT_DATA_DIRS_PROPERTY;
 import static org.assertj.core.api.Assertions.assertThat;
 import static org.junit.Assert.assertEquals;
 
@@ -30,6 +31,7 @@
 import org.junit.ClassRule;
 import org.junit.Rule;
 import org.junit.Test;
+import org.junit.contrib.java.lang.system.RestoreSystemProperties;
 import org.junit.rules.TemporaryFolder;
 
 import org.apache.geode.cache.Region;
@@ -55,6 +57,9 @@ public class ImportDataIntegrationTest {
   @Rule
   public TemporaryFolder tempDir = new TemporaryFolder();
 
+  @Rule
+  public RestoreSystemProperties restoreSystemProperties = new RestoreSystemProperties();
+
   private Region<String, String> region;
   private Path snapshotFile;
   private Path snapshotDir;
@@ -65,6 +70,8 @@ public void setup() throws Exception {
     region = server.getCache().getRegion(TEST_REGION_NAME);
     loadRegion("value");
     Path basePath = tempDir.getRoot().toPath();
+    // configure the test's temporary folder as an export destination
+    System.setProperty(EXPORT_DATA_DIRS_PROPERTY, basePath.toString());
     snapshotFile = basePath.resolve(SNAPSHOT_FILE);
     snapshotDir = basePath.resolve(SNAPSHOT_DIR);
   }

geode-gfsh/src/main/java/org/apache/geode/management/internal/cli/commands/ExportDataCommand.java

@@ -16,6 +16,8 @@
 package org.apache.geode.management.internal.cli.commands;
 
 import java.io.File;
+import java.nio.file.Path;
+import java.nio.file.Paths;
 import java.util.List;
 import java.util.Optional;
 
@@ -33,6 +35,7 @@
 import org.apache.geode.management.internal.cli.result.model.ResultModel;
 import org.apache.geode.management.internal.functions.CliFunctionResult;
 import org.apache.geode.management.internal.i18n.CliStrings;
+import org.apache.geode.security.ResourcePermission;
 import org.apache.geode.security.ResourcePermission.Operation;
 import org.apache.geode.security.ResourcePermission.Resource;
 
@@ -54,6 +57,7 @@ public ResultModel exportData(
           help = CliStrings.EXPORT_DATA__PARALLEL_HELP) boolean parallel) {
 
     authorize(Resource.DATA, Operation.READ, regionName);
+    authorize(Resource.CLUSTER, Operation.WRITE, ResourcePermission.ALL);
     final DistributedMember targetMember = getMember(memberNameOrId);
 
     Optional<ResultModel> validationResult = validatePath(filePath, dirPath, parallel);
@@ -100,6 +104,28 @@ private Optional<ResultModel> validatePath(String filePath, String dirPath, bool
       return Optional.of(ResultModel.createError(CliStrings.format(
           CliStrings.INVALID_FILE_EXTENSION, CliStrings.GEODE_DATA_FILE_EXTENSION)));
     }
+
+    if (filePath != null && containsParentDirectorySegment(filePath)) {
+      return Optional.of(invalidPathError(CliStrings.EXPORT_DATA__FILE, filePath));
+    }
+    if (dirPath != null && containsParentDirectorySegment(dirPath)) {
+      return Optional.of(invalidPathError(CliStrings.EXPORT_DATA__DIR, dirPath));
+    }
+
     return Optional.empty();
   }
+
+  private static boolean containsParentDirectorySegment(String path) {
+    for (Path element : Paths.get(path)) {
+      if ("..".equals(element.toString())) {
+        return true;
+      }
+    }
+    return false;
+  }
+
+  private static ResultModel invalidPathError(String option, String path) {
+    return ResultModel.createError(String.format(
+        "Option \"%s\" must not contain a \"..\" path segment: %s", option, path));
+  }
 }

geode-gfsh/src/main/java/org/apache/geode/management/internal/cli/functions/ExportDataFunction.java

@@ -15,6 +15,9 @@
 package org.apache.geode.management.internal.cli.functions;
 
 import java.io.File;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
 
 import org.apache.geode.cache.Cache;
 import org.apache.geode.cache.Region;
@@ -27,19 +30,34 @@
 import org.apache.geode.management.cli.CliFunction;
 import org.apache.geode.management.internal.functions.CliFunctionResult;
 import org.apache.geode.management.internal.i18n.CliStrings;
+import org.apache.geode.util.internal.GeodeGlossary;
 
 /***
  * Function which carries out the export of a region to a file on a member. Uses the
  * RegionSnapshotService to export the data
  *
- *
+ * <p>
+ * Export destinations are resolved to their canonical form and must be within the export
+ * directories configured for this member.
  */
 public class ExportDataFunction extends CliFunction<String[]> {
   private static final long serialVersionUID = 1L;
 
   private static final String ID =
       "org.apache.geode.management.internal.cli.functions.ExportDataFunction";
 
+  /**
+   * System property naming additional directories this member writes {@code export data} snapshots
+   * into. Several directories may be listed, separated by {@link File#pathSeparator}. Exports into
+   * sub-directories of a configured directory are included.
+   *
+   * <p>
+   * The member's working directory is always configured, since that is where a relative export
+   * path resolves to, so when this property is not set it is the only export destination.
+   */
+  public static final String EXPORT_DATA_DIRS_PROPERTY =
+      GeodeGlossary.GEMFIRE_PREFIX + "export.data.dirs";
+
   @Override
   public String getId() {
     return ID;
@@ -62,7 +80,7 @@ public CliFunctionResult executeFunction(FunctionContext<String[]> context) thro
     String hostName = cache.getDistributedSystem().getDistributedMember().getHost();
     if (region != null) {
       RegionSnapshotService<Object, Object> snapshotService = region.getSnapshotService();
-      final File exportFile = new File(fileName);
+      final File exportFile = resolveExportFile(fileName);
       if (parallel) {
         SnapshotOptions<Object, Object> options = new SnapshotOptionsImpl<>().setParallelMode(true);
         snapshotService.save(exportFile, SnapshotFormat.GEODE, options);
@@ -81,4 +99,42 @@ public CliFunctionResult executeFunction(FunctionContext<String[]> context) thro
 
     return result;
   }
+
+  /**
+   * Resolves the requested export path against the export directories configured for this member.
+   *
+   * @param fileName the path requested by the caller, which may be relative or absolute
+   * @return the canonical file to export to
+   * @throws IllegalArgumentException if the path is not within a configured export directory
+   */
+  static File resolveExportFile(String fileName) throws IOException {
+    File exportFile = new File(fileName).getCanonicalFile();
+    List<File> exportDirs = configuredExportDirs();
+
+    for (File exportDir : exportDirs) {
+      if (exportFile.toPath().startsWith(exportDir.toPath())) {
+        return exportFile;
+      }
+    }
+
+    throw new IllegalArgumentException(String.format(
+        "Cannot export to %s: the path is not within the export directories configured for this member (%s). Use the %s system property to configure additional directories.",
+        exportFile, exportDirs, EXPORT_DATA_DIRS_PROPERTY));
+  }
+
+  private static List<File> configuredExportDirs() throws IOException {
+    List<File> exportDirs = new ArrayList<>();
+    exportDirs.add(new File(System.getProperty("user.dir")).getCanonicalFile());
+
+    String configuredDirs = System.getProperty(EXPORT_DATA_DIRS_PROPERTY);
+    if (configuredDirs != null) {
+      for (String configuredDir : configuredDirs.split(File.pathSeparator)) {
+        if (!configuredDir.trim().isEmpty()) {
+          exportDirs.add(new File(configuredDir.trim()).getCanonicalFile());
+        }
+      }
+    }
+
+    return exportDirs;
+  }
 }

geode-gfsh/src/test/java/org/apache/geode/management/internal/cli/commands/ExportDataCommandTest.java

@@ -15,10 +15,18 @@
 
 package org.apache.geode.management.internal.cli.commands;
 
+import static org.mockito.ArgumentMatchers.any;
+import static org.mockito.ArgumentMatchers.anyString;
+import static org.mockito.Mockito.doNothing;
+import static org.mockito.Mockito.doReturn;
+import static org.mockito.Mockito.mock;
+import static org.mockito.Mockito.spy;
+
 import org.junit.Before;
 import org.junit.ClassRule;
 import org.junit.Test;
 
+import org.apache.geode.distributed.DistributedMember;
 import org.apache.geode.test.junit.rules.GfshParserRule;
 
 
@@ -34,11 +42,35 @@ public void setUp() {
     command = new ExportDataCommand();
   }
 
+  /** A command whose option values are checked without contacting a member. */
+  private ExportDataCommand commandWithMember() {
+    ExportDataCommand withMember = spy(ExportDataCommand.class);
+    doNothing().when(withMember).authorize(any(), any(), anyString());
+    doReturn(mock(DistributedMember.class)).when(withMember).getMember(anyString());
+    return withMember;
+  }
+
   @Test
   public void missingMember() throws Exception {
     // Command parses successfully but fails during execution because cache is null
     gfsh.executeAndAssertThat(command, "export data --region=regionA --file=test")
         .statusIsError()
         .containsOutput("cache");
   }
+
+  @Test
+  public void fileOptionWithParentDirectorySegmentIsRejected() {
+    gfsh.executeAndAssertThat(commandWithMember(),
+        "export data --member=server1 --region=regionA --file=exports/../regionA.gfd")
+        .statusIsError()
+        .containsOutput("must not contain a \"..\" path segment");
+  }
+
+  @Test
+  public void dirOptionWithParentDirectorySegmentIsRejected() {
+    gfsh.executeAndAssertThat(commandWithMember(),
+        "export data --member=server1 --region=regionA --dir=exports/../elsewhere")
+        .statusIsError()
+        .containsOutput("must not contain a \"..\" path segment");
+  }
 }