big-data/hadoop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

YARN-8629. Container cleanup fails while trying to delete Cgroups. (Suma Shivaprasad via wangda)

Browse Source 
Change-Id: I392ef4f8baa84d5d7b1f2e438c560b5426b6d4f2
This commit is contained in:
Wangda Tan 2018-08-07 12:36:55 -07:00
parent b1a59b1644
commit d4258fcad7
1 changed files with 16 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/linux/resources/CGroupsHandlerImpl.java
View File

@ -504,23 +504,29 @@ private void logLineFromTasksFile(File cgf) {
private boolean checkAndDeleteCgroup(File cgf) throws InterruptedException {
boolean deleted = false;
// FileInputStream in = null;
try (FileInputStream in = new FileInputStream(cgf + "/tasks")) {
if (in.read() == -1) {
if ( cgf.exists() ) {
try (FileInputStream in = new FileInputStream(cgf + "/tasks")) {
if (in.read() == -1) {
/*
* "tasks" file is empty, sleep a bit more and then try to delete the
* cgroup. Some versions of linux will occasionally panic due to a race
* condition in this area, hence the paranoia.
*/
Thread.sleep(deleteCGroupDelay);
deleted = cgf.delete();
if (!deleted) {
LOG.warn("Failed attempt to delete cgroup: " + cgf);
Thread.sleep(deleteCGroupDelay);
deleted = cgf.delete();
if (!deleted) {
LOG.warn("Failed attempt to delete cgroup: " + cgf);
}
} else{
logLineFromTasksFile(cgf);
}
} else {
logLineFromTasksFile(cgf);
} catch (IOException e) {
LOG.warn("Failed to read cgroup tasks file. ", e);
}
} catch (IOException e) {
LOG.warn("Failed to read cgroup tasks file. ", e);
} else {
LOG.info("Parent Cgroups directory {} does not exist. Skipping "
+ "deletion", cgf.getPath());
deleted = true;
}
return deleted;
}