[
https://issues.apache.org/jira/browse/CASSANDRA-20448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18054586#comment-18054586
]
Rishabh Saraswat commented on CASSANDRA-20448:
----------------------------------------------
[~smiklosovic] [~bschoeni] I tested the patch (gh pr checkout 3993 -R
apache/cassandra) against the [^script] (make sure to use python version 3.6 -
3.11 before executing the script locally). Got following output
./tools/bin/sstableexpiredblockers -H ks t
WARN 16:34:49,367 Only 19.194GiB free across all data volumes. Consider adding
more capacity to your cluster or removing obsolete snapshots
[BigTableReader:big(path='/home/rishabh/cassandra/data/data/ks/t-1b255f4def2540a60000000000000014/oa-3-big-Data.db')
(minTS = +58043-07-21T05:37:43.338Z, maxTS = +58043-07-21T05:37:43.338Z,
maxLDT = 9223372036854775807, diskSize = 36 bytes)], blocks 1 expired sstables
from getting dropped:
[BigTableReader:big(path='/home/rishabh/cassandra/data/data/ks/t-1b255f4def2540a60000000000000014/oa-4-big-Data.db')
(minTS = +58043-07-26T15:59:12.955Z, maxTS = +58043-07-26T15:59:12.955Z,
maxLDT = 2026-01-27T10:47:41Z, diskSize = 37 bytes)],
[BigTableReader:big(path='/home/rishabh/cassandra/data/data/ks/t-1b255f4def2540a60000000000000014/oa-2-big-Data.db')
(minTS = +58043-07-21T05:37:43.337Z, maxTS = +58043-07-21T05:37:43.337Z,
maxLDT = 9223372036854775807, diskSize = 32 bytes)], blocks 1 expired sstables
from getting dropped:
[BigTableReader:big(path='/home/rishabh/cassandra/data/data/ks/t-1b255f4def2540a60000000000000014/oa-4-big-Data.db')
(minTS = +58043-07-26T15:59:12.955Z, maxTS = +58043-07-26T15:59:12.955Z,
maxLDT = 2026-01-27T10:47:41Z, diskSize = 37 bytes)],
!image-2026-01-27-16-39-03-990.png!
Not so important Note:
I made countless attempts to create blocker using TTL only, but reading
[ExpiredTableBlockers.java
|https://github.com/apache/cassandra/pull/3993/changes#diff-56155fda8f474c54685126b99d5e3aa752ee6e5287064a2732691faed564ceab]
suggested to use clock skew to create historical SSTables with overlapping
timestamp ranges.
Idk but for some reason waiting for 75s didn't work(no blocker SSTables found)
but I waited for more than 75s to get desired output.
> Utility sstableexpiredblockers should report the size of expired tables
> -----------------------------------------------------------------------
>
> Key: CASSANDRA-20448
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20448
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Legacy/Tools
> Reporter: Brad Schoening
> Assignee: Stefan Miklosovic
> Priority: Normal
> Fix For: 5.x
>
> Attachments: image-2026-01-27-16-39-03-990.png, script
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> A key reason to run sstableexpiredblockers is that you have disk space
> accumulating and you wish to investigate what's consuming the disk space.
> If size was included in the output (as shown below), it would make the output
> more useful. Generally, when you find a blocked expired table, the first step
> is to check what size it is to see if it's large enough to be a problem.
> {quote}[BigTableReader(path='/var/lib/cassandra/data/keyspace1/standard1-0665ae80b2d711e886c66d2c86545d91/mc-2-big-Data.db')
> (minTS = 5, maxTS = 5, maxLDT = 2147483647)], blocks 1 expired sstables from
> getting dropped:
> [BigTableReader(path='/var/lib/cassandra/data/keyspace1/standard1-0665ae80b2d711e886c66d2c86545d91/mc-3-big-Data.db')
> (\{*}diskSize = 135.7 GB\{*}, minTS = 1536349775157606, maxTS =
> 1536349780311159, maxLDT = 1536349780)],
> {quote}
> Also, it would be preferable to report min/max timestamp in date-time format
> e.g.:
> {quote} diskSize =134.3 GB, minTimestamp =
> 2025-03-10T00:00:00.000+00:00
> {quote}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
