What’s New in This Release
Introduction
This maintenance release includes many bug fixes, updates, and improvements to GridGain. In addition to the Fixed Issues, this release includes:
-
Improved/augmented messages, monitoring, and logging for various components.
-
Fixed memory leaks, improved performance.
-
Several improvements to Web Console (see Management and Monitoring Changes below).
Changes in Behavior
Detailed Transaction Information Flag
We added a new flag (--tx --info <id>
) to the control.sh|bat
command to get additional information about a specified transaction.
<id>
can be any single transaction identifier in one of several formats. xid
or nearXid
can be passed as GridCacheVersion or UUID.
control.sh --tx --info 2abc7b6aa61-00000000-0a13-93b4-0000-000000000007
control.sh --tx --info GridCacheVersion [topVer=169335904, order=1557855907152, nodeOrder=3]
The information returned includes:
-
Initiator node
-
Transaction parameters: isolation, concurrency, timeout, topology version on which the transaction is/was executed, used caches and groups, transaction states (which can differ across the cluster)
-
Transaction topology (affected nodes):
-
Transaction mappings
-
Each node can handle several mappings where a mapping is a subset of keys affected the by transaction. A mapping can be either:
-
NEAR - local node is a transaction initiator
-
DHT - current keys subset belongs to primary partitions
-
REMOTE - current keys subset belongs to backup partitions
-
-
Dependencies between mappings (e.g. each REMOTE mapping has primary node which owns the same keys in primary partitions)
-
Mapped keys for each mapping and their parameters. Mappings include:
-
Affected keys
-
Their type of access (can be write or read) and their lock ownership status (can be either NO_LOCK - lock is not acquired yet, OWNS_LOCK - key lock is acquired by requested transaction and AWAITS_LOCK - means transaction is waiting for when another transaction will release lock for corresponding key). Example:
Write [lock=OWNS_LOCK]: IgniteTxKey [key=KeyCacheObjectImpl [part=108, val=108, hasValBytes=true], cacheId=1544803905]
-
-
-
This can also work for transactions that have already finished. If the specified transaction persists in the commit/rollback history, all occurrences of the transaction will be printed to output. Example:
Control utility [ver. 2.7.0-SNAPSHOT#20190415-sha1:DEV]
2019 Copyright(C) Apache Software Foundation
User: JUser
Time: 2019-05-15T04:44:48.586
--------------------------------------------------------------------------------
Active transactions not found.
Will try to peek history to find out whether transaction was committed / rolled back.
Transaction was found in completed versions history of the following nodes:
TcpDiscoveryNode [id=b0168b23-ff69-41e6-a044-8ce189800000, addrs=[127.0.0.1], order=1, ver=2.7.0#20190415-sha1:00000000, isClient=false, consistentId=bltTest0]:
State: COMMITTED
TcpDiscoveryNode [id=cc9953a2-2a3e-47c3-83c8-329259900001, addrs=[127.0.0.1], order=2, ver=2.7.0#20190415-sha1:00000000, isClient=false, consistentId=bltTest1]:
State: COMMITTED
...
Known Issue
There’s a compatibility issue between 8.5.8 and version 8.4.4.
Issue GG-18858: The file <dbpath>/<cache_name>/cache_data.dat is not deleted on 8.4.4 on IgniteEx#destroyCache0.
Workaround: Do not execute DROP TABLE queries and do not destroy caches when performing a rolling upgrade from 8.4.4 to 8.5.8.
Installation and Upgrade Information
See the Rolling Upgrades page for information about how to perform automated upgrades and for details about version compatibility.
Fixed Issues
GridGain Community Edition Changes
IGN-13929 |
Cache |
Avoid potential |
IGN-13811 |
Cache |
Fixed missing dependencies in cassandra-store module. |
IGN-13594 |
Cache |
Fixed issue with TTL manager sometimes not cleaning entries from the underlying store. |
IGN-13516 |
Cache |
Fixed: |
IGN-6961 |
Cache |
Fixed hanging parallel write&evict in |
GG-18877 |
Communication |
No thread dumps for tcp-comm-worker thread appear in server logs when client node fails abruptly. |
IGN-13950 |
Control Utilities |
Improved error message when connecting to secured cluster without credentials. |
GG-17426 |
Control Utilities |
Fixed exception when snapshot-utility accesses a file of zero size. |
GG-17344 |
Control Utilities |
Idle-verify: cache-filter 'ALL' will validate system caches, added cache-filter 'USER'. |
IGN-12317 |
Data Structures |
Fixed sporadic |
IGN-13891 |
Deployment |
Fixed an issue with loading different class loaders through p2p. |
IGN-13699 |
Diagnostic |
Added Information about possible long GC pauses during waiting and holding checkpoint write lock. |
IGN-13540 |
Diagnostic |
Improved LRT diagnostic messages: now request from local node is sent to the nearest node to get the dump of a thread that started the transaction. |
IGN-13466 |
Diagnostic |
Added |
IGN-13334 |
Diagnostic |
Added additional logging for client-server connections in TCP-discovery. |
IGN-13276 |
Diagnostic |
Added |
GG-14951 |
Diagnostics & Failure Handling |
Improvements in checking to see if an Exchange worker is alive. |
GG-14950 |
Diagnostics & Failure Handling |
Added configuration parameters for checking to see if system workers are alive. |
GG-14857 |
Diagnostics & Failure Handling |
Added monitoring of critical system workers. |
IGN-13543 |
Discovery |
Fixed infinite 'no next node in topology' loop in |
IGN-13131 |
Discovery |
Grid components of a joining node are initialized with configurations from the grid at the end of join. |
GG-18567 |
Discovery |
Fixed compatibility with old versions. |
IGN-13116 |
Metrics |
Added |
IGN-13260 |
Partition Map Exchange |
Fixed uncaught exception in TTL during partition map exchange. |
GG-17327 |
Partition Map Exchange |
Avoid keeping metrics-related data in exchange history. |
GG-18846 |
Persistent Storage (WAL, Recovery) |
Fixed "Compression of WAL segment was skipped due to unexpected error" spam on node start when WAL compaction enabled. |
GG-18738 |
Persistent Storage (WAL, Recovery) |
Fixed possible data storage corruption on start in no-WAL-archiver mode. |
GG-18736 |
Persistent Storage (WAL, Recovery) |
Introduced |
GG-18860 |
Platform .NET |
Fixed LINQ support for .NET Core 2.1. |
IGN-13941 |
Rebalance |
Fixed node distribution of demanded cache partitions during rebalancing. |
GG-18587 |
Security |
Fix false-positive detection of cluster in not idle state with enabled security. |
GG-17465 |
Security |
Fixed false-positive detection of cluster in not idle state with enabled security. |
GG-14807 |
Security |
Avoid removing expired session from cache on a deactivated cluster. |
IGN-13951 |
SQL |
Fixed memory leak in connection manager. |
IGN-13783 |
SQL |
Fixed case when index rebuild broke consistency between partition and index itself. |
IGN-13704 |
SQL |
Static cache configuration should not override persisted cache configuration. |
GG-18857 |
SQL |
Fixed example |
GG-17434 |
SQL |
Fixed memory leak on unstable topology caused by partition reservation. |
GG-17339 |
SQL |
Fixed memory leak at the JDBC Thin Driver caused when executing any statement. |
GG-17326 |
SQL |
Removed field from internal stored schema properly when the DROP COLUMN statement is executed. |
GG-14963 |
SQL |
Fixed hang on cancel query when the query is executed in |
GG-14873 |
SQL |
Avoid key/val ownership resolution of a field in runtime. |
IGN-13002 |
Tooling |
Added ability to use multiple cache filter options for |
IGN-13792 |
Transactions |
Fixed transaction hang when |
IGN-13624 |
Transactions |
Optimized amount of heap space used for affinity assignment history storing. |
IGN-13286 |
Transactions |
Fixed potential memory leak. |
IGN-12198 |
Transactions |
Provided description for |
GG-17488 |
Transactions |
Fixed memory leak in the |
GridGain Enterprise Edition Changes
GG-17429 |
Data Center Replication |
Fixed deadlock between partition map exchange and the initialization of the Data Center Recovery component. |
GridGain Ultimate Edition Changes
GG-18658 |
Point-in-Time Recovery |
Fixed null pointer exception in logs when server nodes leave cluster during snapshot. |
GG-14777 |
Point-in-Time Recovery |
Fixed: issue with hanging automatic snapshot when system property is set inconsistent in the cluster. |
GG-14760 |
Point-in-Time Recovery |
Fixed possible hang on snapshot operation. |
Management and Monitoring Changes
IGN-12276 |
Web Console |
Improved support for JDBC drivers on project generation. |
GG-18097 |
Web Console |
Added confirmation on changing snapshot schedule enabled state. |
GG-18011 |
Web Console |
Added column with node type: coordinator/server on "Baseline" screen. |
GG-17784 |
Web Console |
Fixed usability issue when cluster has caches with groups. |
GG-18511 |
Web Console |
Added "Pause/Resume/Bootstrap" actions on "Data Center Replication" screen. |
GG-17375 |
Web Console |
Implemented support for Mongo 4.x and latest Ubuntu. |
Related Information
We Value Your Feedback
The GridGain documentation team is focused on constantly improving the product documentation. Your comments and suggestions are always welcome. You can reach us here: docs@gridgain.com
Please visit the documentation for more information.