| __ministry | Hi all, I had problem about senlin that is call many times to keystone and nova api make server to slow and down. Has anyone encountered this error? | 03:50 |
|---|---|---|
| eandersson | __ministry what type of profile and policy are you using? how many clusters? etc | 06:27 |
| __ministry | eandersson: I used os.nova.server. I discovered senlin-health-manager is a process that overloads the system every time it executes health check nova instances. | 07:17 |
| eandersson | What interval are you using? | 07:20 |
| __ministry | my interval is 30s | 07:30 |
| __ministry | i have 104 clusters with 190 nodes | 08:28 |
| __ministry | I had found root cause of this problem. that is when health-manager loading healthcheck, it load more cluster than really. I do grep log had 2784 clusters was loaded health check when current i had 104 clusters | 08:48 |
| eandersson | Interesting | 17:47 |
| eandersson | Is this a Senlin bug __ministry? | 17:47 |
| eandersson | I also wonder if we could make that check more efficient, feels like an O(n) problem | 17:48 |
| eandersson | Is the healthcheck loading clusters that dont exist, or is it loading the same same cluster over and over again? | 17:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!