This message was deleted.
# generals
Slackbot
03/15/2023, 4:17 PMThis message was deleted.
🤔 1
z
Zac Policzer
03/15/2023, 4:19 PMI'm surprised it's health checking if it's gone, that's gotta be a bug. Removing the instance from helix aught to wipe it out or marking the node disabled
e
Enrico Olivelli
03/15/2023, 4:22 PMwhat removes an instance from Helix ?
z
Zac Policzer
03/15/2023, 4:27 PMWell. The node going down itself is intended to be enough to at least stop routing and health checks. In the absence of that there's a helix admin tool that can be leveraged to administer cluster membership (add/remove) or a helix rest API. I need to check if our admin tool wraps this functionality
e
Enrico Olivelli
03/15/2023, 4:27 PMI see.
I guess I still had some old version of the server that advertises the wrong hostname then
Enrico Olivelli
03/15/2023, 4:28 PMso when a venice-server dies the router stops doing the healthcheck because it doesn’t find it anymore on Helix. correct ?
Enrico Olivelli
03/15/2023, 4:29 PM2023-03-15 162915 INFO [BasicClusterDataCache] [Helix-RouterUpdater-event_process-venice-cluster0] Reload InstanceConfig: [venice-server-0.venice-server.mypulsar.svc.cluster.local_54093, venice-server-0_54093]. Takes 6 ms
what does this mean ?
Enrico Olivelli
03/15/2023, 4:30 PMvenice-server-0.venice-server.mypulsar.svc.cluster.local_54093 -> good name
Enrico Olivelli
03/15/2023, 4:30 PMvenice-server-0_54093 -> bad name
Enrico Olivelli
03/15/2023, 4:30 PMI have a patch that allows Venice to advertise the FQDN
Enrico Olivelli
03/15/2023, 4:30 PMbut it seems it is still using the wrong name somewhere
Enrico Olivelli
03/15/2023, 4:32 PMI have found it, sorry for the noise
z
Zac Policzer
03/15/2023, 6:49 PMAh good! What was the gotcha?
e
Enrico Olivelli
03/15/2023, 10:19 PM