Run garbage collection after each symlink and expose --gc in tart prune (#191)

* Properly calculate remotely-retrieved manifest digests

* Implement garbage collection when deleting cached OCI VMs

Sources/tart/Commands/IP.swift

@@ -16,14 +16,21 @@ struct IP: AsyncParsableCommand {
     do {
       let vmDir = try VMStorageLocal().open(name)
       let vmConfig = try VMConfig.init(fromURL: vmDir.configURL)
+      let vmMACAddress = MACAddress(fromString: vmConfig.macAddress.string)!
 
-      guard let ip = try await IP.resolveIP(vmConfig, secondsToWait: wait) else {
+      guard let ipViaDHCP = try await IP.resolveIP(vmMACAddress, secondsToWait: wait) else {
         print("no IP address found, is your VM running?")
 
         Foundation.exit(1)
       }
 
-      print(ip)
+      if let ipViaARP = try ARPCache.ResolveMACAddress(macAddress: vmMACAddress), ipViaARP != ipViaDHCP {
+        fputs("WARNING: DHCP lease and ARP cache entries for MAC address \(vmMACAddress) differ: "
+          + "got \(ipViaDHCP) and \(ipViaARP) respectively, consider reporting this case to"
+          + " https://github.com/cirruslabs/tart/issues/172\n", stderr)
+      }
+
+      print(ipViaDHCP)
 
       Foundation.exit(0)
     } catch {
@@ -33,12 +40,11 @@ struct IP: AsyncParsableCommand {
     }
   }
 
-  static public func resolveIP(_ config: VMConfig, secondsToWait: UInt16) async throws -> IPv4Address? {
+  static public func resolveIP(_ vmMACAddress: MACAddress, secondsToWait: UInt16) async throws -> IPv4Address? {
     let waitUntil = Calendar.current.date(byAdding: .second, value: Int(secondsToWait), to: Date.now)!
-    let vmMacAddress = MACAddress(fromString: config.macAddress.string)!
 
     repeat {
-      if let ip = try Leases().resolveMACAddress(macAddress: vmMacAddress) {
+      if let ip = try Leases().resolveMACAddress(macAddress: vmMACAddress) {
         return ip
       }
 

Sources/tart/Commands/Prune.swift

@@ -16,6 +16,9 @@ struct Prune: AsyncParsableCommand {
     valueName: "n"))
   var cacheBudget: UInt?
 
+  @Flag(help: .hidden)
+  var gc: Bool = false
+
   func validate() throws {
     if olderThan == nil && cacheBudget == nil {
       throw ValidationError("at least one criteria must be specified")
@@ -24,6 +27,10 @@ struct Prune: AsyncParsableCommand {
 
   func run() async throws {
     do {
+      if gc {
+        try VMStorageOCI().gc()
+      }
+
       // Clean up cache entries based on last accessed date
       if let olderThan = olderThan {
         let olderThanInterval = Int(exactly: olderThan)!.days.timeInterval

Sources/tart/MACAddressResolver/ARPCache.swift

@@ -0,0 +1,119 @@
+import Foundation
+import Network
+import Virtualization
+
+struct ARPCommandFailedError: Error, CustomStringConvertible {
+    var terminationReason: Process.TerminationReason
+    var terminationStatus: Int32
+
+    var description: String {
+        var reason: String
+
+        switch terminationReason {
+        case .exit:
+            reason = "exit code \(terminationStatus)"
+        case .uncaughtSignal:
+            reason = "uncaught signal"
+        default:
+            reason = "unknown reason"
+        }
+
+        return "arp command failed: \(reason)"
+    }
+}
+
+struct ARPCommandYieldedInvalidOutputError: Error, CustomStringConvertible {
+    var explanation: String
+
+    var description: String {
+        "arp command yielded invalid output: \(explanation)"
+    }
+}
+
+struct ARPCacheInternalError: Error, CustomStringConvertible {
+    var explanation: String
+
+    var description: String {
+        "ARPCache internal error: \(explanation)"
+    }
+}
+
+struct ARPCache {
+    static func ResolveMACAddress(macAddress: MACAddress, bridgeOnly: Bool = true) throws -> IPv4Address? {
+        let process = Process.init()
+        process.executableURL = URL.init(fileURLWithPath: "/usr/sbin/arp")
+        process.arguments = ["-an"]
+
+        let pipe = Pipe()
+        process.standardOutput = pipe
+        process.standardError = pipe
+        process.standardInput = FileHandle.nullDevice
+
+        try process.run()
+        process.waitUntilExit()
+
+        if !(process.terminationReason == .exit && process.terminationStatus == 0) {
+            throw ARPCommandFailedError(
+                    terminationReason: process.terminationReason,
+                    terminationStatus: process.terminationStatus)
+        }
+
+        guard let rawLines = try pipe.fileHandleForReading.readToEnd() else {
+            throw ARPCommandYieldedInvalidOutputError(explanation: "empty output")
+        }
+        let lines = String(decoding: rawLines, as: UTF8.self)
+                .trimmingCharacters(in: .whitespacesAndNewlines)
+                .components(separatedBy: "\n")
+
+        // Based on https://opensource.apple.com/source/network_cmds/network_cmds-606.40.2/arp.tproj/arp.c.auto.html
+        let regex = try NSRegularExpression(pattern: #"^.* \((?<ip>.*)\) at (?<mac>.*) on (?<interface>.*) .*$"#)
+
+        for line in lines {
+            let nsLineRange = NSRange(line.startIndex..<line.endIndex, in: line)
+
+            guard let match = regex.firstMatch(in: line, range: nsLineRange) else {
+                throw ARPCommandYieldedInvalidOutputError(explanation: "unparseable entry \"\(line)\"")
+            }
+
+            let rawIP = try match.getCaptureGroup(name: "ip", for: line)
+            guard let ip = IPv4Address(rawIP) else {
+                throw ARPCommandYieldedInvalidOutputError(explanation: "failed to parse IPv4 address \(rawIP)")
+            }
+
+            let rawMAC = try match.getCaptureGroup(name: "mac", for: line)
+            if rawMAC == "(incomplete)" {
+                continue
+            }
+            guard let mac = MACAddress(fromString: rawMAC) else {
+                throw ARPCommandYieldedInvalidOutputError(explanation: "failed to parse MAC address \(rawMAC)")
+            }
+
+            let interface = try match.getCaptureGroup(name: "interface", for: line)
+            if bridgeOnly && !interface.starts(with: "bridge") {
+                continue
+            }
+
+            if macAddress == mac {
+                return ip
+            }
+        }
+
+        return nil
+    }
+}
+
+extension NSTextCheckingResult {
+    func getCaptureGroup(name: String, for string: String) throws -> String {
+        let nsRange = self.range(withName: name)
+
+        if nsRange.location == NSNotFound {
+            throw ARPCacheInternalError(explanation: "attempted to retrieve non-existent named capture group \(name)")
+        }
+
+        guard let range = Range.init(nsRange, in: string) else {
+            throw ARPCacheInternalError(explanation: "failed to convert NSRange to Range")
+        }
+
+        return String(string[range])
+    }
+}

Sources/tart/MACAddressResolver/MACAddress.swift

@@ -16,6 +16,6 @@ struct MACAddress: Equatable, Hashable, CustomStringConvertible {
   }
 
   var description: String {
-    return String(format: "%02x:%02x:%02x:%02x:%02x:%02x", mac[0], mac[1], mac[2], mac[3], mac[4], mac[5])
+    String(format: "%02x:%02x:%02x:%02x:%02x:%02x", mac[0], mac[1], mac[2], mac[3], mac[4], mac[5])
   }
 }

Sources/tart/URL+AccessDate.swift

@@ -20,6 +20,6 @@ extension URL {
 
 extension Date {
   func asTimeval() -> timeval {
-    timeval(tv_sec: timeIntervalSince1970.toUnit(.second)!, tv_usec: 0)
+    timeval(tv_sec: Int(timeIntervalSince1970), tv_usec: 0)
   }
 }

Sources/tart/VMDirectory.swift

@@ -20,6 +20,10 @@ struct VMDirectory: Prunable {
     baseURL.appendingPathComponent("nvram.bin")
   }
 
+  var explicitlyPulledMark: URL {
+    baseURL.appendingPathComponent(".explicitly-pulled")
+  }
+
   var name: String {
     baseURL.lastPathComponent
   }
@@ -89,4 +93,12 @@ struct VMDirectory: Prunable {
   func sizeBytes() throws -> Int {
     try configURL.sizeBytes() + diskURL.sizeBytes() + nvramURL.sizeBytes()
   }
+
+  func markExplicitlyPulled() {
+    FileManager.default.createFile(atPath: explicitlyPulledMark.path, contents: nil)
+  }
+
+  func isExplicitlyPulled() -> Bool {
+    FileManager.default.fileExists(atPath: explicitlyPulledMark.path)
+  }
 }

Sources/tart/VMStorageOCI.swift

@@ -42,6 +42,44 @@ class VMStorageOCI: PrunableStorage {
 
   func delete(_ name: RemoteName) throws {
     try FileManager.default.removeItem(at: vmURL(name))
+    try gc()
+  }
+
+  func gc() throws {
+    var refCounts = Dictionary<URL, UInt>()
+
+    guard let enumerator = FileManager.default.enumerator(at: baseURL,
+            includingPropertiesForKeys: [.isSymbolicLinkKey]) else {
+      return
+    }
+
+    for case let foundURL as URL in enumerator {
+      let isSymlink = try foundURL.resourceValues(forKeys: [.isSymbolicLinkKey]).isSymbolicLink!
+
+      // Perform garbage collection for tag-based images
+      // with broken outgoing references
+      if isSymlink && foundURL == foundURL.resolvingSymlinksInPath() {
+        try FileManager.default.removeItem(at: foundURL)
+        continue
+      }
+
+      let vmDir = VMDirectory(baseURL: foundURL.resolvingSymlinksInPath())
+      if !vmDir.initialized {
+        continue
+      }
+
+      refCounts[vmDir.baseURL] = (refCounts[vmDir.baseURL] ?? 0) + (isSymlink ? 1 : 0)
+    }
+
+    // Perform garbage collection for digest-based images
+    // with no incoming references
+    for (baseURL, incRefCount) in refCounts {
+      let vmDir = VMDirectory(baseURL: baseURL)
+
+      if !vmDir.isExplicitlyPulled() && incRefCount == 0 {
+        try FileManager.default.removeItem(at: baseURL)
+      }
+    }
   }
 
   func list() throws -> [(String, VMDirectory, Bool)] {
@@ -82,10 +120,10 @@ class VMStorageOCI: PrunableStorage {
   func pull(_ name: RemoteName, registry: Registry) async throws {
     defaultLogger.appendNewLine("pulling manifest...")
 
-    let (manifest, _) = try await registry.pullManifest(reference: name.reference.value)
+    let (manifest, manifestData) = try await registry.pullManifest(reference: name.reference.value)
 
-    var digestName = RemoteName(host: name.host, namespace: name.namespace,
-            reference: Reference(digest: try manifest.digest()))
+    let digestName = RemoteName(host: name.host, namespace: name.namespace,
+            reference: Reference(digest: Digest.hash(manifestData)))
 
     if !exists(digestName) {
       let tmpVMDir = try VMDirectory.temporary()
@@ -113,8 +151,12 @@ class VMStorageOCI: PrunableStorage {
     }
 
     if name != digestName {
-      // Overwrite the old symbolic link
+      // Create new or overwrite the old symbolic link
       try link(from: digestName, to: name)
+    } else {
+      // Ensure that images pulled by content digest
+      // are excluded from garbage collection
+      VMDirectory(baseURL: vmURL(name)).markExplicitlyPulled()
     }
   }
 
@@ -124,6 +166,8 @@ class VMStorageOCI: PrunableStorage {
     }
 
     try FileManager.default.createSymbolicLink(at: vmURL(to), withDestinationURL: vmURL(from))
+
+    try gc()
   }
 }
 

Sources/tart/VNC/ScreenSharingVNC.swift

@@ -10,7 +10,8 @@ class ScreenSharingVNC: VNC {
   }
 
   func waitForURL() async throws -> URL {
-    let ip = try await IP.resolveIP(vmConfig, secondsToWait: 60)
+    let vmMACAddress = MACAddress(fromString: vmConfig.macAddress.string)!
+    let ip = try await IP.resolveIP(vmMACAddress, secondsToWait: 60)
 
     if let ip = ip {
       return URL(string: "vnc://\(ip)")!