[
https://issues.apache.org/jira/browse/CASSANDRA-21073?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmitry Konstantinov updated CASSANDRA-21073:
--------------------------------------------
Description:
Currently "active compaction remaining time" is calculated as
{code:java}
(total - completed) / compaction_throughput{code}
the real processing speed can be much lower that compaction_throughput
configuration limit, so the value can be highly inaccurate.
We can use some realtime information (rate or the time since a start of a
compaction) to make the reported value more accurate.
Option 1:
{code:java}
(total - completed) / current compaction throughput (1 minute){code}
Option 2: add compaction start time or time since start to CompactionStats
{code:java}
(total - completed) / (completed / time since compaction start){code}
Like here, I set compaction_throughput to 1024MiB (just as an example):
{code:java}
./bin/nodetool compactionstats -H
concurrent compactors 1
pending tasks 2
stress stress_table 2
compactions completed 10
data compacted 14.82 GiB
compactions aborted 0
compactions reduced 0
sstables dropped from compaction 0
15 minute rate 24.88/minute
mean rate 100.63/hour
compaction throughput (MiB/s) 1024.0
current compaction throughput (1 minute) 47.052 MiB/s
current compaction throughput (5 minute) 25.934 MiB/s
current compaction throughput (15 minute) 10.843 MiB/s
id compaction type keyspace table
completed total unit progress
c3b5c9d0-d6b9-11f0-b467-b1b09c3e4b2e Compaction stress stress_table
288.84 MiB 1.36 GiB bytes 20.74%
active compaction remaining time 0h00m01s {code}
was:
Currently "active compaction remaining time" is calculated as
{code:java}
(total - completed)/ compaction_throughput{code}
the real processing speed can be much lower that compaction_throughput
configuration limit, so the value can be highly inaccurate.
We can use some realtime information (rate or the time since a start of a
compaction) to make the reported value more accurate.
Like here, I set compaction_throughput to 1024MiB (just as an example):
{code:java}
./bin/nodetool compactionstats -H
concurrent compactors 1
pending tasks 2
stress stress_table 2
compactions completed 10
data compacted 14.82 GiB
compactions aborted 0
compactions reduced 0
sstables dropped from compaction 0
15 minute rate 24.88/minute
mean rate 100.63/hour
compaction throughput (MiB/s) 1024.0
current compaction throughput (1 minute) 47.052 MiB/s
current compaction throughput (5 minute) 25.934 MiB/s
current compaction throughput (15 minute) 10.843 MiB/s
id compaction type keyspace table
completed total unit progress
c3b5c9d0-d6b9-11f0-b467-b1b09c3e4b2e Compaction stress stress_table
288.84 MiB 1.36 GiB bytes 20.74%
active compaction remaining time 0h00m01s {code}
> Make "active compaction remaining time" reported by nodetool compactionstats
> more accurate
> ------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-21073
> URL: https://issues.apache.org/jira/browse/CASSANDRA-21073
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Local/Compaction, Tool/nodetool
> Reporter: Dmitry Konstantinov
> Priority: Normal
>
> Currently "active compaction remaining time" is calculated as
> {code:java}
> (total - completed) / compaction_throughput{code}
> the real processing speed can be much lower that compaction_throughput
> configuration limit, so the value can be highly inaccurate.
> We can use some realtime information (rate or the time since a start of a
> compaction) to make the reported value more accurate.
> Option 1:
> {code:java}
> (total - completed) / current compaction throughput (1 minute){code}
> Option 2: add compaction start time or time since start to CompactionStats
> {code:java}
> (total - completed) / (completed / time since compaction start){code}
> Like here, I set compaction_throughput to 1024MiB (just as an example):
> {code:java}
> ./bin/nodetool compactionstats -H
> concurrent compactors 1
>
> pending tasks 2
>
> stress stress_table 2
>
> compactions completed 10
>
> data compacted 14.82 GiB
>
> compactions aborted 0
>
> compactions reduced 0
>
> sstables dropped from compaction 0
>
> 15 minute rate 24.88/minute
>
> mean rate 100.63/hour
>
> compaction throughput (MiB/s) 1024.0
>
> current compaction throughput (1 minute) 47.052 MiB/s
>
> current compaction throughput (5 minute) 25.934 MiB/s
>
> current compaction throughput (15 minute) 10.843 MiB/s
>
> id compaction type keyspace table
> completed total unit progress
> c3b5c9d0-d6b9-11f0-b467-b1b09c3e4b2e Compaction stress
> stress_table 288.84 MiB 1.36 GiB bytes 20.74%
> active compaction remaining time 0h00m01s {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
