-
Notifications
You must be signed in to change notification settings - Fork 3.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rpc for querying ingesters to get chunk ids from its store #2601
rpc for querying ingesters to get chunk ids from its store #2601
Conversation
b932b0f
to
9fd20fd
Compare
e61058a
to
a21b78b
Compare
Codecov Report
@@ Coverage Diff @@
## master #2601 +/- ##
==========================================
- Coverage 62.87% 61.41% -1.47%
==========================================
Files 170 172 +2
Lines 15051 13312 -1739
==========================================
- Hits 9464 8175 -1289
+ Misses 4827 4394 -433
+ Partials 760 743 -17
|
9303b3a
to
c45fa56
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Minor editing and format suggestions.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this looks good @sandeepsukhani, some feedback from @oddlittlebird to address but lets get this merged!
found something i think we need to fix
3810578
to
0317f58
Compare
…h chunk store and ingesters
…db-shipper or query store is true
1. query chunk-refs parallely from store and ingesters. 2. limit look back for chunk refs and labels from ingesters to start of schema when using boltdb-shipper.
…ngester or querier
66149e1
to
2946f52
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks great @sandeepsukhani nice work!
What this PR does / why we need it:
When using boltdb shipper ingesters have to serve logs from the store as well to include logs from chunks which are flushed but its index might not be uploaded yet. Due to this extra workload, ingesters are OOMing when someone runs a huge query.
To avoid this we need to offload the workload of logs processing to queriers. This can be done by adding a new RPC to ingesters to let queriers query for IDs of the chunks which were already flushed to the store.
This is still a draft PR to discuss the RPC definition. The actual implementation would follow.
Checklist