Timeline for Replacing a dead 3.11 node fails with OOM
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 28 at 17:19 | comment | added | Mario | Secondary indexes in Cassandra can create a large number of small SSTables, which exacerbates the problem during repairs and replacements. How many of them do you still have in your cluster? We usually avoid using them. Can you drop them or we should try to fix the cluster with the secondary index? | |
| Jun 19 at 14:02 | comment | added | John Koyle | I added some additional info above. I suspect that if I assassinate the node and add a new one that would get me back to 6 in the cluster. However, I suspect initiating repairs on the keyspace with the secondary indexes will lead to the same issue I'm having with bootstrapping a replacement. Thoughts? | |
| Jun 17 at 18:18 | history | edited | mustaccio | CC BY-SA 4.0 |
added 57 characters in body
|
| Jun 17 at 17:37 | history | edited | Mario | CC BY-SA 4.0 |
added 133 characters in body
|
| S Jun 17 at 17:33 | review | First answers | |||
| Jun 17 at 18:18 | |||||
| S Jun 17 at 17:33 | history | answered | Mario | CC BY-SA 4.0 |