[
https://issues.apache.org/jira/browse/TS-721?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13014277#comment-13014277
]
Leif Hedstrom commented on TS-721:
----------------------------------
So, looking at the very old code (it was never used), I see
// Old stats
varSetInt("proxy.node.http.cache_total_hits", hits);
varSetInt("proxy.node.http.cache_total_misses", miss);
varSetFloat("proxy.node.http.cache_hit_ratio", hitRate);
// new stats
varSetInt("proxy.node.cache_total_hits", hits);
varSetInt("proxy.node.cache_total_misses", miss);
varSetFloat("proxy.node.cache_hit_ratio", hitRate);
meaning, the old stats are "old". So, they were never ported to the new stats
processor (which uses the XML configs). Do we really need to support these old
stats? I don't see them used anywhere, so not sure how it's affecting the
cluster stats?
My recommendation would be that we remove these old ones from RecordsConfig.cc,
to avoid confusion.
> incorrect http hit ratio in stats
> ---------------------------------
>
> Key: TS-721
> URL: https://issues.apache.org/jira/browse/TS-721
> Project: Traffic Server
> Issue Type: Bug
> Components: HTTP, Stats
> Affects Versions: 2.1.7
> Reporter: Zhao Yongming
> Assignee: Leif Hedstrom
> Fix For: 2.1.8
>
>
> in my production, the proxy.node.http.cache_hit_ratio is always 0, it should
> be the same as proxy.node.cache_hit_ratio, that make my cluster wide stats
> error too:
> [root@cache190 ~]# lynx -source http://localhost:81/stat/ | grep ratio
> proxy.config.cluster.cluster_configuration=cluster.config
> proxy.config.cluster.cluster_load_clear_duration=24
> proxy.config.cluster.cluster_load_exceed_duration=4
> proxy.config.icp.icp_configuration=icp.config
> proxy.config.update.update_configuration=update.config
> proxy.node.http.cache_hit_ratio=0.000000
> proxy.node.http.cache_hit_ratio_int_pct=0
> proxy.node.http.bandwidth_hit_ratio=0.904612
> proxy.node.http.bandwidth_hit_ratio_int_pct=90
> proxy.node.bandwidth_hit_ratio=0.904612
> proxy.node.bandwidth_hit_ratio_int_pct=90
> proxy.node.hostdb.hit_ratio=0.049913
> proxy.node.hostdb.hit_ratio_int_pct=5
> proxy.node.cache_hit_ratio=0.883374
> proxy.node.cache_hit_ratio_int_pct=88
> proxy.node.cache_hit_ratio_avg_10s_int_pct=93
> proxy.node.bandwidth_hit_ratio_avg_10s_int_pct=93
> proxy.node.bandwidth_hit_ratio_avg_10s=0.934583
> proxy.node.cache_hit_ratio_avg_10s=0.930153
> proxy.node.hostdb.hit_ratio_avg_10s=0.996169
> proxy.cluster.bandwidth_hit_ratio=0.898589
> proxy.cluster.bandwidth_hit_ratio_int_pct=1023958400
> proxy.cluster.bandwidth_hit_ratio_avg_10s=0.938777
> proxy.cluster.http.cache_hit_ratio=0.900161
> proxy.cluster.http.cache_hit_ratio_int_pct=1063552512
> proxy.cluster.http.bandwidth_hit_ratio=0.915798
> proxy.cluster.http.bandwidth_hit_ratio_int_pct=1063552576
> proxy.cluster.cache_hit_ratio=0.900161
> proxy.cluster.cache_hit_ratio_int_pct=180
> proxy.cluster.cache_hit_ratio_avg_10s=0.935070
> proxy.cluster.hostdb.hit_ratio=0.047426
> proxy.cluster.hostdb.hit_ratio_int_pct=6
> proxy.cluster.hostdb.hit_ratio_avg_10s=0.337434
> proxy.process.cluster.configuration_changes=1
> [root@cache190 ~]# traffic_line -r proxy.node.http.cache_hit_ratio
> 0.000000
> in my situation:
> proxy.node.http.cache_hit_ratio=0.000000
> proxy.node.http.cache_hit_ratio_int_pct=0
> should be same as:
> proxy.node.cache_hit_ratio=0.883374
> proxy.node.cache_hit_ratio_int_pct=88
> there is some strange error in cluster too:
> proxy.cluster.bandwidth_hit_ratio_int_pct=1023958400
> proxy.cluster.http.cache_hit_ratio_int_pct=1063552512
> proxy.cluster.http.bandwidth_hit_ratio_int_pct=1063552576
> proxy.cluster.cache_hit_ratio_int_pct=180
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
