Pivotal Knowledge Base

Follow

Index Creation in GemFire fails with a distributed deadlock

Index Creation in GemFire fails with a distributed deadlock

Symptoms

  • A deadlock may occur during region initialization when index creation happens locally or through a remote request in vFabric GemFire
  • In the GemFire log files, you see that each member is waiting for a disk store hosted at an empty host list:

    [info 2012/03/29 17:48:00.282 EDT  GemFireServer tid=0xa0] Region /SomeRegion, bucket 111 initialized with data from /192.169.1.100:/path/to/data/store/SomeRegion created at timestamp 1332442223821 version 0 diskStoreId 263ba267-d1d5-4abe-9af7-cbb11f09e23a is waiting for the data previously hosted at [] to be available

Cause

This issue occurs in GemFire 6.6.x and earlier versions when a partitioned region with indexes begins recovering data from a persistence file that may not be the most recent.

Resolution

To resolve this issue, upgrade to GemFire 7.0 or later.
 
To download GemFire 7.0 or later, see the Pivotal Network

Comments

Powered by Zendesk