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

HDFS-15629. Add seqno when warning slow mirror/disk in BlockReceiver. Contributed by Haibin Huang.

Browse Source
This commit is contained in:
Ayush Saxena 2020-10-19 10:34:17 +05:30
parent 3a0922fb42
commit 173310e2f5
1 changed files with 15 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BlockReceiver.java
View File

@ -409,7 +409,7 @@ boolean packetSentInTime() {
* Flush block data and metadata files to disk.
* @throws IOException
*/
void flushOrSync(boolean isSync) throws IOException {
void flushOrSync(boolean isSync, long seqno) throws IOException {
long flushTotalNanos = 0;
long begin = Time.monotonicNow();
if (checksumOut != null) {
@ -448,7 +448,8 @@ void flushOrSync(boolean isSync) throws IOException {
LOG.warn("Slow flushOrSync took " + duration + "ms (threshold="
+ datanodeSlowLogThresholdMs + "ms), isSync:" + isSync + ", flushTotalNanos="
+ flushTotalNanos + "ns, volume=" + getVolumeBaseUri()
+ ", blockId=" + replicaInfo.getBlockId());
+ ", blockId=" + replicaInfo.getBlockId()
+ ", seqno=" + seqno);
}
}
@ -606,7 +607,8 @@ private int receivePacket() throws IOException {
LOG.warn("Slow BlockReceiver write packet to mirror took " + duration
+ "ms (threshold=" + datanodeSlowLogThresholdMs + "ms), "
+ "downstream DNs=" + Arrays.toString(downstreamDNs)
+ ", blockId=" + replicaInfo.getBlockId());
+ ", blockId=" + replicaInfo.getBlockId()
+ ", seqno=" + seqno);
}
} catch (IOException e) {
handleMirrorOutError(e);
@ -622,7 +624,7 @@ private int receivePacket() throws IOException {
}
// sync block if requested
if (syncBlock) {
flushOrSync(true);
flushOrSync(true, seqno);
}
} else {
final int checksumLen = diskChecksum.getChecksumSize(len);
@ -741,7 +743,8 @@ private int receivePacket() throws IOException {
LOG.warn("Slow BlockReceiver write data to disk cost:" + duration
+ "ms (threshold=" + datanodeSlowLogThresholdMs + "ms), "
+ "volume=" + getVolumeBaseUri()
+ ", blockId=" + replicaInfo.getBlockId());
+ ", blockId=" + replicaInfo.getBlockId()
+ ", seqno=" + seqno);
}
if (duration > maxWriteToDiskMs) {
@ -814,14 +817,14 @@ private int receivePacket() throws IOException {
}
/// flush entire packet, sync if requested
flushOrSync(syncBlock);
flushOrSync(syncBlock, seqno);
replicaInfo.setLastChecksumAndDataLen(offsetInBlock, lastCrc);
datanode.metrics.incrBytesWritten(len);
datanode.metrics.incrTotalWriteTime(duration);
manageWriterOsCache(offsetInBlock);
manageWriterOsCache(offsetInBlock, seqno);
}
} catch (IOException iex) {
// Volume error check moved to FileIoProvider
@ -887,7 +890,7 @@ private static byte[] copyLastChunkChecksum(byte[] array, int size, int end) {
return Arrays.copyOfRange(array, end - size, end);
}
private void manageWriterOsCache(long offsetInBlock) {
private void manageWriterOsCache(long offsetInBlock, long seqno) {
try {
if (streams.getOutFd() != null &&
offsetInBlock > lastCacheManagementOffset + CACHE_DROP_LAG_BYTES) {
@ -934,7 +937,8 @@ private void manageWriterOsCache(long offsetInBlock) {
LOG.warn("Slow manageWriterOsCache took " + duration
+ "ms (threshold=" + datanodeSlowLogThresholdMs
+ "ms), volume=" + getVolumeBaseUri()
+ ", blockId=" + replicaInfo.getBlockId());
+ ", blockId=" + replicaInfo.getBlockId()
+ ", seqno=" + seqno);
}
}
} catch (Throwable t) {
@ -1646,7 +1650,8 @@ private void sendAckUpstreamUnprotected(PipelineAck ack, long seqno,
+ "ms (threshold=" + datanodeSlowLogThresholdMs + "ms), " + myString
+ ", replyAck=" + replyAck
+ ", downstream DNs=" + Arrays.toString(downstreamDNs)
+ ", blockId=" + replicaInfo.getBlockId());
+ ", blockId=" + replicaInfo.getBlockId()
+ ", seqno=" + seqno);
} else if (LOG.isDebugEnabled()) {
LOG.debug(myString + ", replyAck=" + replyAck);
}