Limits names lookback in Elasticsearch #1538
Conversation
This reuses `QUERY_LOOKBACK` to limit names queries (service and span names) from searching all indexes. This is particularly important for high-volume sites (ex 100% sampling or high traffic), and those who do not use curator or otherwise to prune data. Fixes #1462 (at least helps a lot with no schema change)
|
circleci is freaking out on license header dates (which was a problem we had in travis prior) |
| @Value("${zipkin.storage.strict-trace-id:true}") boolean strictTraceId) { | ||
| return elasticsearch.toBuilder(client).strictTraceId(strictTraceId); | ||
| @Value("${zipkin.storage.strict-trace-id:true}") boolean strictTraceId, | ||
| @Value("${zipkin.query.lookback:86400000}") int namesLookback) { |
There was a problem hiding this comment.
ps I'm ok to add this as a elasticsearch property instead. ex zipkin.storage.elasticsearch.names-lookback (we have similar things in cassandra)
There was a problem hiding this comment.
Add this as a elasticsearch property may be a better idea. zipkin.query.lookback is a 32bit int variable which has a maximum of 2_147_483_647(ms) that is about 59.6621 hours. It is really not enough.
A zipkin.storage.elasticsearch.names-lookback property in secs or even days can be great. Or may be you can use a 64bit variable for zipkin.query.lookback.
ps. Should this be a new issue?
|
sounds like a bug. mind sending a PR to fix this (along with a unit test to
show it working)?
nice find!
|
|
@adriancole Thank you for your conform, I will check if there are still similar issues and try fixing them. |
This reuses
QUERY_LOOKBACKto limit names queries (service and spannames) from searching all indexes. This is particularly important for
high-volume sites (ex 100% sampling or high traffic), and those who do
not use curator or otherwise to prune data.
Fixes #1462 (at least helps a lot with no schema change)