Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[CELEBORN-1646] Catch exception of Files#getFileStore for DeviceMonitor and StorageManager for input/ouput error #2809

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+56 −35
Open

[CELEBORN-1646] Catch exception of Files#getFileStore for DeviceMonitor and StorageManager for input/ouput error #2809

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 8 additions & 1 deletion worker/src/main/scala/org/apache/celeborn/service/deploy/worker/storage/DeviceMonitor.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,14 @@ class LocalDeviceMonitor(
.groupBy(_.deviceInfo)
.foreach { case (deviceInfo: DeviceInfo, diskInfos: List[DiskInfo]) =>
val deviceLabel = Map("device" -> deviceInfo.name)
def usage: DeviceMonitor.DiskUsageInfo = DeviceMonitor.getDiskUsageInfos(diskInfos.head)
def usage: DeviceMonitor.DiskUsageInfo =
try {
DeviceMonitor.getDiskUsageInfos(diskInfos.head)
} catch {
case t: Throwable =>
logError("Device monitor get usage infos failed.", t)
DeviceMonitor.DiskUsageInfo(0L, 0L, 0L, 0)
}
workerSource.addGauge(WorkerSource.DEVICE_OS_TOTAL_CAPACITY, deviceLabel) { () =>
usage.totalSpace
}
Expand Down
82 changes: 48 additions & 34 deletions worker/src/main/scala/org/apache/celeborn/service/deploy/worker/storage/StorageManager.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -727,15 +727,19 @@ final private[worker] class StorageManager(conf: CelebornConf, workerSource: Abs
val appIds = shuffleKeySet().asScala.map(key => Utils.splitShuffleKey(key)._1)
while (retryTimes < conf.workerCheckFileCleanMaxRetries) {
val localCleaned =
!disksSnapshot().filter(_.status != DiskStatus.IO_HANG).exists { diskInfo =>
diskInfo.dirs.exists {
case workingDir if workingDir.exists() =>
// Don't check appDirs that store information in the fileInfos
workingDir.listFiles().exists(appDir => !appIds.contains(appDir.getName))
case _ =>
false
!disksSnapshot()
.filter(diskInfo =>
diskInfo.status != DiskStatus.IO_HANG || diskInfo.status != DiskStatus.READ_OR_WRITE_FAILURE)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there any problem about this? why need filter READ_OR_WRITE_FAILURE?

.exists {
diskInfo =>
diskInfo.dirs.exists {
case workingDir if workingDir.exists() =>
// Don't check appDirs that store information in the fileInfos
workingDir.listFiles().exists(appDir => !appIds.contains(appDir.getName))
case _ =>
false
}
}
}

val dfsCleaned = hadoopFs match {
case dfs: FileSystem =>
Expand Down Expand Up @@ -853,33 +857,43 @@ final private[worker] class StorageManager(conf: CelebornConf, workerSource: Abs
}

def updateDiskInfos(): Unit = this.synchronized {
disksSnapshot().filter(_.status != DiskStatus.IO_HANG).foreach { diskInfo =>
val totalUsage = diskInfo.dirs.map { dir =>
val writers = workingDirWriters.get(dir)
if (writers != null) {
writers.synchronized {
writers.values.asScala.map(_.getDiskFileInfo.getFileLength).sum
disksSnapshot()
.filter(diskInfo =>
diskInfo.status != DiskStatus.IO_HANG || diskInfo.status != DiskStatus.READ_OR_WRITE_FAILURE)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need to filter CRITICAL_ERROR as well?

.foreach {
diskInfo =>
val totalUsage = diskInfo.dirs.map { dir =>
val writers = workingDirWriters.get(dir)
if (writers != null) {
writers.synchronized {
writers.values.asScala.map(_.getDiskFileInfo.getFileLength).sum
}
} else {
0
}
}.sum

try {
val (fileSystemReportedUsableSpace, fileSystemReportedTotalSpace) =
getFileSystemReportedSpace(diskInfo.mountPoint)
val workingDirUsableSpace =
Math.min(diskInfo.configuredUsableSpace - totalUsage, fileSystemReportedUsableSpace)
val minimumReserveSize =
DiskUtils.getMinimumUsableSize(diskInfo, diskReserveSize, diskReserveRatio)
val usableSpace = Math.max(workingDirUsableSpace - minimumReserveSize, 0)
logDebug(
s"Update diskInfo:${diskInfo.mountPoint} workingDirUsableSpace:$workingDirUsableSpace fileMeta:$fileSystemReportedUsableSpace" +
s"conf:${diskInfo.configuredUsableSpace} totalUsage:$totalUsage totalSpace:$fileSystemReportedTotalSpace" +
s"minimumReserveSize:$minimumReserveSize usableSpace:$usableSpace")
diskInfo.setUsableSpace(usableSpace)
diskInfo.setTotalSpace(fileSystemReportedTotalSpace)
} catch {
case t: Throwable =>
logError(s"Update diskInfo:${diskInfo.mountPoint} failed.", t)
}
} else {
0
}
}.sum

val (fileSystemReportedUsableSpace, fileSystemReportedTotalSpace) =
getFileSystemReportedSpace(diskInfo.mountPoint)
val workingDirUsableSpace =
Math.min(diskInfo.configuredUsableSpace - totalUsage, fileSystemReportedUsableSpace)
val minimumReserveSize =
DiskUtils.getMinimumUsableSize(diskInfo, diskReserveSize, diskReserveRatio)
val usableSpace = Math.max(workingDirUsableSpace - minimumReserveSize, 0)
logDebug(s"updateDiskInfos workingDirUsableSpace:$workingDirUsableSpace filemeta:$fileSystemReportedUsableSpace" +
s"conf:${diskInfo.configuredUsableSpace} totalUsage:$totalUsage totalSpace:$fileSystemReportedTotalSpace" +
s"minimumReserveSize:$minimumReserveSize usableSpace:$usableSpace")
diskInfo.setUsableSpace(usableSpace)
diskInfo.setTotalSpace(fileSystemReportedTotalSpace)
diskInfo.updateFlushTime()
diskInfo.updateFetchTime()
}
diskInfo.updateFlushTime()
diskInfo.updateFetchTime()
}
logInfo(s"Updated diskInfos:\n${disksSnapshot().mkString("\n")}")
}

Expand Down
Loading