[Linux-cluster] Crashing machines with luci and ricci
Dirk H. Schulz
dirk.schulz at kinzesberg.de
Wed Jul 2 18:55:17 UTC 2008
I have tried setting up a cluster with ricci and luci. I have done the
- set up 2 cluster nodes
- current patch level applied (5.2)
- installed ricci on the nodes and luci on a management station
- used luci web interface to setup the cluster
After initial setup luci stated that one node could not be reached or had
ricci not running. Both nodes were set up identical and could be reached
fine. ricci was running on both machines.
So I used the "restart the cluster" button - and that crashed both nodes
within 10 minutes. One machine was unreachable nearly at once, the other
had 100 % CPU load for several minutes before going down.
Now so far there is not much I could have done wrong (at least not
according to documentation). So I would like to know: Is this normal? Is
using ricci and luci a bad idea because they simply do not work?
Or the other way round: Are folks out there using these tools with positive
results - and are there nuts and bolts I could have avoided?
Any hint or help is appreciated.
More information about the Linux-cluster