[
https://issues.apache.org/jira/browse/YARN-2637?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14253048#comment-14253048
]
Wangda Tan commented on YARN-2637:
----------------------------------
Hi [~cwelch],
bq. Regarding "FiCaSchedulerApp constructor: Tests were depending on the
previous, incorrect behavior to run ...",
I think that is fine if we have to change a lot of tests to avod such null
checking.
bq. Tests were depending on the previous, incorrect behavior to run
I think there should at least one AM can be launched in each queue, otherwise
the queue totally makes no sense. Imaging a user can see a queue and cluster
has available resource, but app in the queue is still pending. I think we can
fix the issue together in this ticket, or file a separate JIRA, I prefer
previous one.
And in addition, we should see how many tests will fail instead of set value
here. We should not add hard code such configuration in MockRM's constructor, a
developer may write unit test like
{code}
Configuration conf = new Configuration(..);
conf.set(AM_RESOURCE, 0.123f);
MockRM rm = new MockRM(conf);
{code}
Because the AM_RESOURCE value will be overwritten in new MockRM's logic.
bq. -re maximumActiveApplications - this is a good question. Before this change
it was possible to effectively set this value by just doing a bit of math
because the "pretend" AM size was a fixed value.
I just thought it is not correct, as you mentioned, we shouldn't have logic
depend on this incorrect behavior, it is not a backward compatible change to
me, it is just leave some necessary logic in implementation/configuration.
If a user wants to specify #app in a queue just doing tests, set a proper
AM_PERCENTAGE and also launches AM with fixed capacity will be a very easy way,
do I under-estimate this problem?
Thanks,
Wangda
> maximum-am-resource-percent could be violated when resource of AM is >
> minimumAllocation
> ----------------------------------------------------------------------------------------
>
> Key: YARN-2637
> URL: https://issues.apache.org/jira/browse/YARN-2637
> Project: Hadoop YARN
> Issue Type: Bug
> Components: resourcemanager
> Affects Versions: 2.6.0
> Reporter: Wangda Tan
> Assignee: Craig Welch
> Priority: Critical
> Attachments: YARN-2637.0.patch, YARN-2637.1.patch,
> YARN-2637.12.patch, YARN-2637.13.patch, YARN-2637.15.patch,
> YARN-2637.16.patch, YARN-2637.17.patch, YARN-2637.18.patch,
> YARN-2637.19.patch, YARN-2637.2.patch, YARN-2637.6.patch, YARN-2637.7.patch,
> YARN-2637.9.patch
>
>
> Currently, number of AM in leaf queue will be calculated in following way:
> {code}
> max_am_resource = queue_max_capacity * maximum_am_resource_percent
> #max_am_number = max_am_resource / minimum_allocation
> #max_am_number_for_each_user = #max_am_number * userlimit * userlimit_factor
> {code}
> And when submit new application to RM, it will check if an app can be
> activated in following way:
> {code}
> for (Iterator<FiCaSchedulerApp> i=pendingApplications.iterator();
> i.hasNext(); ) {
> FiCaSchedulerApp application = i.next();
>
> // Check queue limit
> if (getNumActiveApplications() >= getMaximumActiveApplications()) {
> break;
> }
>
> // Check user limit
> User user = getUser(application.getUser());
> if (user.getActiveApplications() <
> getMaximumActiveApplicationsPerUser()) {
> user.activateApplication();
> activeApplications.add(application);
> i.remove();
> LOG.info("Application " + application.getApplicationId() +
> " from user: " + application.getUser() +
> " activated in queue: " + getQueueName());
> }
> }
> {code}
> An example is,
> If a queue has capacity = 1G, max_am_resource_percent = 0.2, the maximum
> resource that AM can use is 200M, assuming minimum_allocation=1M, #am can be
> launched is 200, and if user uses 5M for each AM (> minimum_allocation). All
> apps can still be activated, and it will occupy all resource of a queue
> instead of only a max_am_resource_percent of a queue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
