Details
-
Bug
-
Resolution: Fixed
-
Critical
-
Cheshire-Cat
-
CB EE 7.0.0-5127
-
Untriaged
-
Centos 64-bit
-
-
1
-
Unknown
Description
Summary:
The RAM usage on indexer node (172.23.106.238) is almost 12GiB even when there are no gsi indexes. I had many indexes on collections, and then I dropped all collections (and hence all indexes got dropped). I waited close to 24 hours and the RAM usage still didn't drop.
Here's the RAM usage on indexer node after waiting for 24 hours after all indexes got dropped:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
|
13304 couchba+ 20 0 15.8g 12.4g 19168 S 2.7 52.9 424:39.43 indexer
|
106880 couchba+ 20 0 4743884 227704 3692 S 2.0 0.9 322:27.12 beam.smp
|
107810 couchba+ 20 0 1280224 239304 13836 S 1.3 1.0 23:52.86 prometheus
|
Steps to Reproduce:
1. Create 2kv, 1 n1ql, 14 index nodes cluster
2. Create a bucket with 1K collections, and a total of 1K items
3. Create 10 indexes per collection (so a total of 10K indexes that are deferred)
4. Build 6K indexes from step 3
5. Flush bucket, and drop all collections. So total index count is 0 here.
6. Rebalance-out all index nodes except one index node (172.23.106.238)
Wait for sufficiently long and observe that RAM usage by indexer on the above node doesn't drop as mentioned in the summary above
Attaching logs from the resulting 4 node cluster at step 6
Attachments
Issue Links
- is duplicated by
-
MB-46241 Service 'indexer' exited with status 137
- Closed