Vladimir Malinovskiy created IGNITE-15029:
---------------------------------------------
Summary: Add additional messages in a coordinator when
distribution isn't ideal
Key: IGNITE-15029
URL: https://issues.apache.org/jira/browse/IGNITE-15029
Project: Ignite
Issue Type: Improvement
Reporter: Vladimir Malinovskiy
In time of rebalance the cluster has a temporary distribution, that have not
primary nodes in the own places. That mechanism also is called late affinity
assignment.
In this time coordinator waiting of rebalance on all nodes in topology before
it will change a distribution to ideal (by affinity function opinion). Until
that will happen need to tell user abut the distribution is not ideal with a
list of nodes (caches, groups, partitions) which it waits.
All information which can necessary for the message is contained here:
_CacheAffinitySharedManager#waitInfo_
I would want to see this message periodically (2-6 ones a minute) until a
partitions' distribution won't become the ideal.
Update: After discussion it was decided not to print each partition Id to avoid
getting too long messages. Also rebalance statistics was added to local node
metrics that are printed to log.
Example of a message:
{{Current affinity assignment is not ideal, it is waiting for cache:
grp=[grpId=-2100569601, nodes=[node=[id=e5d9354a-d909-4bbb-bd92-b70212f00001,
partsNum=100], node=[id=23315c26-24ab-4b50-ac83-fdd64ca00000, partsNum=100],
node=[id=39a954c1-c97d-4b7d-8405-c27554600002, partsNum=100]]],
grp=[grpId=-1049814975, nodes=[node=[id=39a954c1-c97d-4b7d-8405-c27554600002,
partsNum=347]]], grp=[grpId=-1049814974,
nodes=[node=[id=39a954c1-c97d-4b7d-8405-c27554600002, partsNum=347]]]}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
