You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@tez.apache.org by "Hitesh Shah (JIRA)" <ji...@apache.org> on 2015/09/04 01:24:46 UTC
[jira] [Commented] (TEZ-2774) Reduce logging in the AM
[ https://issues.apache.org/jira/browse/TEZ-2774?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14730019#comment-14730019 ]
Hitesh Shah commented on TEZ-2774:
----------------------------------
Comments:
{code}
LOG.info("Comparing client version with AM version"
+ ", clientVersion=" + clientVersion
+ ", AMVersion=" + dagVersionInfo.getVersion());
{code}
- this has been useful at times when debugging rolling upgrade goof-ups. One-time line per AM instance so would be useful to retain.
{code}
if (LOG.isDebugEnabled()) {
1089 LOG.debug("remoteTaskSpec:" + remoteTaskSpec);
1090 }
{code}
- Does this need any additional context on what is being logged? i.e. taskAttemptId or taskId?
{code}
} else if (vertex.completedTaskCount %50 == 0 || vertex.completedTaskCount == vertex.tasks.size()) {
2082 LOG.info("Checking tasks for vertex completion for "
2083 + vertex.logIdentifier
{code}
- should this be "modulo max(50, numTasks/50)" i.e log this at max 50 times?
{code}
for (ContainerStatus status : statuses) {
412 LOG.debug("Container " + status.getContainerId() + " is completed");
413 }
{code}
- should enhance to log container exit code, diagnostics etc.
- maybe log failed containers at info ( not only in this case but others that were converted to debug )?
- not required if AM is shutting down - which is the case if stopStarted is true
{code}
LOG.debug("No taskRequests. Container's idle timeout delay expired or is new. " +
640 "Releasing container"
641 + ", containerId=" + heldContainer.getContainer().getId()
642 + ", containerExpiryTime="
643 + heldContainer.getContainerExpiryTime()
644 + ", idleTimeout=" + idleContainerTimeoutMin
645 + ", taskRequestsCount=" + taskRequests.size()
646 + ", heldContainers=" + heldContainers.size()
647 + ", delayedContainers=" + delayedContainerManager.delayedContainers.size()
648 + ", isNew=" + isNew);
649 }
639 releaseUnassignedContainers( 650
{code}
- might be worth retaining if this is not a very frequent log. If frequent, then yes we can remove.
{code}
if (LOG.isDebugEnabled()) {
810 LOG.debug("Releasing held container as either there are pending but "
811 + " unmatched requests or this is not a session"
812 + ", containerId=" + heldContainer.container.getId()
813 + ", pendingTasks=" + taskRequests.size()
814 + ", isSession=" + getContext().isSession()
815 + ". isNew=" + isNew);
816 }
{code}
- might be worth retaining with a less heavy message.
typo: timelineAclsClas - change to TimelineACLManagerClass
{code}
if (isEnabled) {
82 allocatorClassName = conf.get(TezConfiguration.TEZ_TASK_SCALE_MEMORY_ALLOCATOR_CLASS,
83 TezConfiguration.TEZ_TASK_SCALE_MEMORY_ALLOCATOR_CLASS_DEFAULT);
84 } else {
85 allocatorClassName = null;
86 }
{code}
- non logging change?
{code}
private final long LOG_INTERVAL = 30000l;
{code}
- from 2 seconds to 30 seconds?
Tez runtime log changes are present. Maybe change jira title?
> Reduce logging in the AM
> ------------------------
>
> Key: TEZ-2774
> URL: https://issues.apache.org/jira/browse/TEZ-2774
> Project: Apache Tez
> Issue Type: Sub-task
> Reporter: Siddharth Seth
> Assignee: Siddharth Seth
> Attachments: TEZ-2774.1.txt
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)