[jira] [Commented] (CASSANDRA-14668) Diag events for read repairs

2018-08-31 Thread mck (JIRA)


[ 
https://issues.apache.org/jira/browse/CASSANDRA-14668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16598791#comment-16598791
 ] 

mck commented on CASSANDRA-14668:
-

latest…


|| branch || testall || dtest ||
| 
[CASSANDRA-14668|https://github.com/spodkowinski/cassandra/tree/CASSANDRA-14668]
  | 
[!https://circleci.com/gh/spodkowinski/cassandra/tree/CASSANDRA-14668.svg?style=svg!|https://circleci.com/gh/spodkowinski/cassandra/tree/CASSANDRA-14668]
 | 
[!https://builds.apache.org/view/A-D/view/Cassandra/job/Cassandra-devbranch-dtest/624/badge/icon!|https://builds.apache.org/view/A-D/view/Cassandra/job/Cassandra-devbranch-dtest/624/]
 |

I'm +1 on the code.

In a system that has a badly stale node (causing many/most reads to RR) this 
will create a lot of diagnostics events (when enabled), but that's the whole 
point of being able to enable just specific diagnostic events.



> Diag events for read repairs
> 
>
> Key: CASSANDRA-14668
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14668
> Project: Cassandra
>  Issue Type: Improvement
>  Components: Observability
>Reporter: Stefan Podkowinski
>Assignee: Stefan Podkowinski
>Priority: Major
> Fix For: 4.x
>
>
> Read repairs have been a highly discussed topic during the last months and 
> also saw some significant code changes. I'd like to be better prepared in 
> case we need to investigate any further RR issues in the future, by adding 
> diagnostic events that can be enabled for exposing informations such as:
>  * contacted endpoints
>  * digest responses by endpoint
>  * affected partition keys
>  * speculated reads / writes
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org



[jira] [Commented] (CASSANDRA-14668) Diag events for read repairs

2018-08-31 Thread Stefan Podkowinski (JIRA)


[ 
https://issues.apache.org/jira/browse/CASSANDRA-14668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16598377#comment-16598377
 ] 

Stefan Podkowinski commented on CASSANDRA-14668:


Example events from a single read repair as retrieved via JMX.

ReadRepairEvent:
{noformat}
class=org.apache.cassandra.service.reads.repair.ReadRepairEvent, 
type=START_REPAIR, 
command=SELECT * FROM ks.table1 WHERE key = a LIMIT 100, 
consistency=ALL,
speculativeRetry=PERCENTILE, 
keyspace=ks, 
table=table1, 
allEndpoints=[127.0.0.1:7000, 127.0.0.2:7000], 
endpointDestinations=[127.0.0.1:7000, 127.0.0.2:7000], 
thread=Native-Transport-Requests-1, 
digestsByEndpoint={
127.0.0.2:7000={digestHex=d41d8cd98f00b204e9800998ecf8427e, 
isDigestResponse=true}, 
127.0.0.1:7000={digestHex=82577faad6f8c6450786e68df35db686, 
isDigestResponse=false}
}, 
ts=1535700434509
{noformat}
PartitionRepairEvent:
{noformat}
class=org.apache.cassandra.service.reads.repair.PartitionRepairEvent, 
type=SEND_INITIAL_REPAIRS, 
consistency=ALL, 
keyspace=ks, 
mutation=Mutation(keyspace='ks', key='61', modifications=[
  [ks.table1] key=a partition_deletion=deletedAt=-9223372036854775808, 
localDeletion=2147483647 columns=[[] | [c1 v1]]
Row[info=[ts=1535700208937182] ]: EMPTY | [c1=1 ts=1535700208937182], [v1=1 
ts=1535700208937182]
]), 
destination=127.0.0.2:7000, 
thread=Native-Transport-Requests-1, 
key=61, 
token=-8839064797231613815, 
ts=1535700434522
{noformat}

> Diag events for read repairs
> 
>
> Key: CASSANDRA-14668
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14668
> Project: Cassandra
>  Issue Type: Improvement
>  Components: Observability
>Reporter: Stefan Podkowinski
>Assignee: Stefan Podkowinski
>Priority: Major
> Fix For: 4.x
>
>
> Read repairs have been a highly discussed topic during the last months and 
> also saw some significant code changes. I'd like to be better prepared in 
> case we need to investigate any further RR issues in the future, by adding 
> diagnostic events that can be enabled for exposing informations such as:
>  * contacted endpoints
>  * digest responses by endpoint
>  * affected partition keys
>  * speculated reads / writes
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org



[jira] [Commented] (CASSANDRA-14668) Diag events for read repairs

2018-08-30 Thread mck (JIRA)


[ 
https://issues.apache.org/jira/browse/CASSANDRA-14668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16598250#comment-16598250
 ] 

mck commented on CASSANDRA-14668:
-

builds.apache.org came back up today, so kick off the dtests for this…

|| branch || testall || dtest ||
| [WIP-14668|spodkowinski/cassandra/tree/WIP-14668] | 
[!https://circleci.com/gh/spodkowinski/cassandra/tree/WIP-14668.svg?style=svg!|https://circleci.com/gh/spodkowinski/cassandra/tree/WIP-14668]
 | 
[!https://builds.apache.org/view/A-D/view/Cassandra/job/Cassandra-devbranch-dtest/621/badge/icon!|https://builds.apache.org/view/A-D/view/Cassandra/job/Cassandra-devbranch-dtest/621/]
 |

> Diag events for read repairs
> 
>
> Key: CASSANDRA-14668
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14668
> Project: Cassandra
>  Issue Type: Improvement
>  Components: Observability
>Reporter: Stefan Podkowinski
>Assignee: Stefan Podkowinski
>Priority: Major
> Fix For: 4.x
>
>
> Read repairs have been a highly discussed topic during the last months and 
> also saw some significant code changes. I'd like to be better prepared in 
> case we need to investigate any further RR issues in the future, by adding 
> diagnostic events that can be enabled for exposing informations such as:
>  * contacted endpoints
>  * digest responses by endpoint
>  * affected partition keys
>  * speculated reads / writes
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org