[
https://issues.apache.org/jira/browse/MESOS-2373?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14333107#comment-14333107
]
Michael Park commented on MESOS-2373:
-------------------------------------
[~bmahler]: As part of the changes to the allocator to store {{hashmap<SlaveID,
Resources}} instead of {{Resources}}, {{allocator::addFramework(FrameworkID,
FrameworkInfo, Resources)}}} needed to be updated to
{{allocator::addFramework(FrameworkID, FrameworkInfo, hashmap<SlaveID,
Resources>)}} as well. This leads to the change in
{{master::Framework::usedResources}} and
{{master::Framework::offeredResources}} from {{Resources}} to
{{hashmap<SlaveID, Resources>}}. This is very much desired in terms of
collecting accurate information (since frameworks can be offered resources from
multiple slaves), but it does lead to a breaking API change since both
{{usedResources}} and {{offeredResources}} are exposed through the HTTP API (as
you mentioned offline).
I'm not sure what the expectation is for this kind of situation. Is it possible
for us to change the API and make sure to document it in the release notes +
upgrades doc?
> DRFSorter needs to distinguish resources from different slaves.
> ---------------------------------------------------------------
>
> Key: MESOS-2373
> URL: https://issues.apache.org/jira/browse/MESOS-2373
> Project: Mesos
> Issue Type: Bug
> Components: allocation
> Reporter: Michael Park
> Assignee: Michael Park
> Labels: mesosphere
>
> Currently the {{DRFSorter}} aggregates total and allocated resources across
> multiple slaves, which only works for scalar resources. We need to
> distinguish resources from different slaves.
> Suppose we have 2 slaves and 1 framework. The framework is allocated all
> resources from both slaves.
> {code}
> Resources slaveResources =
> Resources::parse("cpus:2;mem:512;ports:[31000-32000]").get();
> DRFSorter sorter;
> sorter.add(slaveResources); // Add slave1 resources
> sorter.add(slaveResources); // Add slave2 resources
> // Total resources in sorter at this point is
> // cpus(*):4; mem(*):1024; ports(*):[31000-32000].
> // The scalar resources get aggregated correctly but ports do not.
> sorter.add("F");
> // The 2 calls to allocated only works because we simply do:
> // allocation[name] += resources;
> // without checking that the 'resources' is available in the total.
> sorter.allocated("F", slaveResources);
> sorter.allocated("F", slaveResources);
> // At this point, sorter.allocation("F") is:
> // cpus(*):4; mem(*):1024; ports(*):[31000-32000].
> {code}
> To provide some context, this issue came up while trying to reserve all
> unreserved resources from every offer.
> {code}
> for (const Offer& offer : offers) {
> Resources unreserved = offer.resources().unreserved();
> Resources reserved = unreserved.flatten(role, Resource::FRAMEWORK);
> Offer::Operation reserve;
> reserve.set_type(Offer::Operation::RESERVE);
> reserve.mutable_reserve()->mutable_resources()->CopyFrom(reserved);
>
> driver->acceptOffers({offer.id()}, {reserve});
> }
> {code}
> Suppose the slave resources are the same as above:
> {quote}
> Slave1: {{cpus(\*):2; mem(\*):512; ports(\*):\[31000-32000\]}}
> Slave2: {{cpus(\*):2; mem(\*):512; ports(\*):\[31000-32000\]}}
> {quote}
> Initial (incorrect) total resources in the DRFSorter is:
> {quote}
> {{cpus(\*):4; mem(\*):1024; ports(\*):\[31000-32000\]}}
> {quote}
> We receive 2 offers, 1 from each slave:
> {quote}
> Offer1: {{cpus(\*):2; mem(\*):512; ports(\*):\[31000-32000\]}}
> Offer2: {{cpus(\*):2; mem(\*):512; ports(\*):\[31000-32000\]}}
> {quote}
> At this point, the resources allocated for the framework is:
> {quote}
> {{cpus(\*):4; mem(\*):1024; ports(\*):\[31000-32000\]}}
> {quote}
> After first {{RESERVE}} operation with Offer1:
> The allocated resources for the framework becomes:
> {quote}
> {{cpus(\*):2; mem(\*):512; cpus(role):2; mem(role):512;
> ports(role):\[31000-32000\]}}
> {quote}
> During second {{RESERVE}} operation with Offer2:
> {code:title=HierarchicalAllocatorProcess::updateAllocation}
> // ...
> FrameworkSorter* frameworkSorter =
> frameworkSorters[frameworks\[frameworkId\].role];
> Resources allocation = frameworkSorter->allocation(frameworkId.value());
> // Update the allocated resources.
> Try<Resources> updatedAllocation = allocation.apply(operations);
> CHECK_SOME(updatedAllocation);
> // ...
> {code}
> {{allocation}} in the above code is:
> {quote}
> {{cpus(\*):2; mem(\*):512; cpus(role):2; mem(role):512;
> ports(role):\[31000-32000\]}}
> {quote}
> We try to {{apply}} a {{RESERVE}} operation and we fail to find
> {{ports(\*):\[31000-32000\]}} which leads to the {{CHECK}} fail at
> {{CHECK_SOME(updatedAllocation);}}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
