You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by John Wong <go...@gmail.com> on 2015/10/03 22:59:30 UTC
Manually expire OpsCenter metric data in cassandra
Hi.
I am using Cassandra 1.2.15 and OpsCenter 5.0.2 (non-enterprise). We are
planning to upgrade in the upcoming week.
I found some old opscenter tables still lurking around.
/mnt/data/OpsCenter# du -h
68K ./events_timeline
18M ./rollups86400
172K ./pdps
269M ./rollups7200
68K ./events
4.0K ./bestpractice_results
228K ./settings
5.0G ./rollups300
28G ./rollups60
34G .
and if we look at rollups60
/mnt/data/OpsCenter/rollups60# ll
total 29195188
drwxr-xr-x 2 cassandra cassandra 20480 Oct 3 19:59 ./
drwxr-xr-x 11 cassandra cassandra 4096 Jun 13 04:54 ../
-rw-r--r-- 1 cassandra cassandra 1064886 Jul 15 01:33
OpsCenter-rollups60-ic-212-CompressionInfo.db
-rw-r--r-- 1 cassandra cassandra 2980474664 Jul 15 01:33
OpsCenter-rollups60-ic-212-Data.db
-rw-r--r-- 1 cassandra cassandra 1776 Jul 15 01:33
OpsCenter-rollups60-ic-212-Filter.db
-rw-r--r-- 1 cassandra cassandra 2002960 Jul 15 01:33
OpsCenter-rollups60-ic-212-Index.db
....
-rw-r--r-- 1 cassandra cassandra 5706306736 Oct 3 20:42
OpsCenter-rollups60-tmp-ic-672-Data.db
-rw-r--r-- 1 cassandra cassandra 3670016 Oct 3 20:42
OpsCenter-rollups60-tmp-ic-672-Index.db
But we have been using the expire technique as described in
http://docs.datastax.com/en/opscenter/5.1/opsc/configure/opscChangingPerformanceDataExpiration_t.html#opsc_changing_performance_data_expiration_times_t__clusternameconf_unique_1
.
Our config is expire 2 hour after 28 days.
[cassandra_metrics]
1min_ttl = 86400
5min_ttl = 604800
2hr_ttl = 2419200
Looking at one of the rollup tables:
cqlsh:OpsCenter> select * from rollups300 limit 10;
key |
column1 | value
--------------------------------------------------------------------+-----------+----------------------------
10.0.1.1-SOME_CF-notifications-getRecentBloomFilterFalseRatio | 703578847
| 0x000000000000000000000000
10.0.1.1-SOME_CF-getRecentBloomFilterFalseRatio | 703579147 |
0x000000000000000000000000
....
Here is my TTL query:
cqlsh:OpsCenter> select TTL (value) from rollups300 limit 2;
ttl(value)
------------
604500
604194
I can see TTL is going down each time I query.
How do I expire these rolllup data myself? Or to be exactly, why do I still
have data from July? Also, how can I confirm the TTL is actually set and
working?
I am okay with truncating the entire keyspace once but I'd like to know if
there are more granular options in case this occurs again.
Thanks.
John