Red Hat Bugzilla – Bug 52338
nanny erroneously causes fail-over of healty active node in fos cluster
Last modified: 2007-04-18 12:36:13 EDT
Description of Problem: Version-Release number of selected component (if applicable): Piranha v. 0.4.17-2 How Reproducible: Sometimes Steps to Reproduce: 1. Install fos cluster in busy network 2. Wait, sometimes days 3. Observe fail-over Actual Results: Fail-over service works, but for wrong reason Expected Results: Spurios fail-overs due to connection error to alive service should not occur Additional Information: It should be quite easy to add a variable to lvs.cf in the services sections, where the nanny is allowed top make a given number of connection attempts before deciding that the service in question on the active node is dead. This exactly like the heartbeat works.
There is a much more up to date version of piranha available. Please use these and get back to me if you are experiancing the same problems ftp://ftp.linux.org/uk/pub/linux/piranha you need to install scsi_reserve, ipvsadm and piranha in that order additionally you'll notice two branchs 6.2 and 7.x, be aware that the ipvsadm packages are specifit to the kernel version the 6.2 branch is for 2.2 kernels and the 7.X for 2.4 NOTE: the web interface has changed since 0.4 you will have to start the interface using service piranha-gui start you will have to setup a passwd fir the piranha user piranha-passwd <a passwd> and you will have to connect to the interface using http://localhost:3636/ (username piranha passwd <a passwd> as above) Phil =--=
Can we close this bug due to lack of feedback?
Yeap! CLOSED!