[jira] [Commented] (CASSANDRA-8660) Cassandra makes Java 1.7+ SEGFAULT

["Pierre Belanger (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-8660?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14284991#comment-14284991 ] 

Pierre Belanger commented on CASSANDRA-8660:
--------------------------------------------

FYI

- Out of 525 cassandra crashes , all hs_err file reports current thread was CompactionExecutor.
  # grep ^= hs_err_pid* | grep CompactionExecutor
  525

- We have the following for our CF :

  compaction_strategy = 'LeveledCompactionStrategy'
  sstable_size_in_mb = 200
  sstable_compression = 'SnappyCompressor'
  tombstone_compaction_interval = 300
  tombstone_threshold = 0.1;

- cassandra.yaml stripped :

cluster_name: 'Default Cluster'
hinted_handoff_enabled: true
max_hint_window_in_ms: 10800000 # 3 hours
hinted_handoff_throttle_in_kb: 1024
max_hints_delivery_threads: 2
batchlog_replay_throttle_in_kb: 1024
authenticator: AllowAllAuthenticator
authorizer: AllowAllAuthorizer
permissions_validity_in_ms: 2000
partitioner: org.apache.cassandra.dht.Murmur3Partitioner
data_file_directories:
    - /home/var/cassandra/data
commitlog_directory: /home/var/cassandra/commitlog
disk_failure_policy: stop
commit_failure_policy: stop
key_cache_size_in_mb: 600
key_cache_save_period: 14400
row_cache_size_in_mb: 20000
row_cache_save_period: 14400
counter_cache_size_in_mb: 0
counter_cache_save_period: 7200
memory_allocator: JEMallocAllocator
saved_caches_directory: /home/var/cassandra/saved_caches
commitlog_sync: periodic
commitlog_sync_period_in_ms: 10000
commitlog_segment_size_in_mb: 32
seed_provider:
    - class_name: com.company.cassandra.SmartrAutoSeedProvider
      parameters:
          - numHosts: "3"
concurrent_reads: 160
concurrent_writes: 160
concurrent_counter_writes: 32
memtable_heap_space_in_mb: 6000
memtable_offheap_space_in_mb: 6000
memtable_allocation_type: heap_buffers
commitlog_total_space_in_mb: 8192
index_summary_capacity_in_mb:
index_summary_resize_interval_in_minutes: 60
trickle_fsync: true
trickle_fsync_interval_in_kb: 10240
storage_port: 7000
ssl_storage_port: 7001
start_native_transport: true
native_transport_port: 9042
native_transport_max_threads: 1024
start_rpc: true
rpc_port: 9160
rpc_keepalive: true
rpc_server_type: sync
thrift_framed_transport_size_in_mb: 15
incremental_backups: false
snapshot_before_compaction: false
auto_snapshot: false
tombstone_warn_threshold: 1000
tombstone_failure_threshold: 1000000
column_index_size_in_kb: 128
batch_size_warn_threshold_in_kb: 5
concurrent_compactors: 40
compaction_throughput_mb_per_sec: 0
sstable_preemptive_open_interval_in_mb: 50
read_request_timeout_in_ms: 60000
range_request_timeout_in_ms: 60000
write_request_timeout_in_ms: 60000
counter_write_request_timeout_in_ms: 5000
cas_contention_timeout_in_ms: 1000
truncate_request_timeout_in_ms: 60000
request_timeout_in_ms: 30000
cross_node_timeout: false
endpoint_snitch: SimpleSnitch
dynamic_snitch_update_interval_in_ms: 100
dynamic_snitch_reset_interval_in_ms: 600000
dynamic_snitch_badness_threshold: 0.1
request_scheduler: org.apache.cassandra.scheduler.NoScheduler
internode_compression: dc
inter_dc_tcp_nodelay: false
disk_access_mode: mmap_index_only
thrift_max_message_length_in_mb: 16
index_interval: 128

> Cassandra makes Java 1.7+ SEGFAULT
> ----------------------------------
>
>                 Key: CASSANDRA-8660
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8660
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>         Environment: Linux Redhat 6.5
>            Reporter: Pierre Belanger
>            Priority: Critical
>         Attachments: hs_err_pid12004.log
>
>
> We upgraded a 3 nodes cluster from Cassandra 2.0.11-5f54285e9e to 2.1-7e6d9eb842.
> With Cassandra 2.1 , within ~10 min.  cassandra manages to makes Java 1.7.0_72-b14 segfault.  This is also tested and same behavior w/ Java 1.7.0_60-b19 and 1.8.0_25-b17.
> We were able to use same data, downgrade back to 2.0 and things ran for ~3-4 days w/ no hick up.   Upgrading back to Cassandra-2.1, Java does a segfault again.
> See attached hs_err file.
> Pierre



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)