CentOS Bug Tracker
CentOS Website

Viewing Issue Simple Details Jump to Notes ] View Advanced ] Issue History ] Print ]
ID Category Severity Reproducibility Date Submitted Last Update
0003924 [CentOS-5] piranha major always 2009-10-22 13:30 2009-11-05 04:34
Reporter moylo View Status public  
Assigned To
Priority normal Resolution open  
Status new   Product Version 5.4
Summary 0003924: piranha-0.8.4-13.el5 breaks LVS
Description The piranha-0.8.4-13.el5 package that was part of the 5.4 update breaks Director LVS (I have not tried the update with NAT LVS).

/var/log/messages complains that:
nanny[19636]: /sbin/ipvsadm command failed!

In a typical MASTER/SECONDARY LVS server setup 'pulse' on the SECONDARY node will not automatically failover when this issue occurs on the MASTER (tested on a SECONDARY box with working LVS)

The workaround is to downgrade to piranha-0.8.4-11.el5.


Additional Information Linux dlvs3 2.6.18-164.el5 0000001 SMP Thu Sep 3 03:28:30 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux
[root@dlvs3 ~]#

[root@dlvs3 ~]# cat /etc/redhat-release
CentOS release 5.4 (Final)

[root@dlvs3 ~]# getenforce
Disabled
Tags No tags attached.
Attached Files

- Relationships

-  Notes
(0010109)
moylo (reporter)
2009-10-22 16:50

Forgot to add. ipvsadm -ln output on a defective system returns 0 server mappings

[root@lvs3 ~]# ipvsadm -ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port Forward Weight ActiveConn InActConn
(0010230)
fabrana (reporter)
2009-11-01 19:47

Similar problem on 32bit release
It happens immediately,when pulse starts

...
Nov 1 19:25:22 loadbalancer02 lvs[4641]: nanny died! shutting down lvs
Nov 1 19:25:22 loadbalancer02 last message repeated 9 times
Nov 1 19:25:22 loadbalancer02 lvs[4641]: shutting down virtual service
...

CentOS release 5.4 (Final)
Linux 2.6.18-164.2.1.el5 0000001 SMP Wed Sep 30 12:55:19 EDT 2009 i686 i686 i386 GNU/Linux
Network type: Direct Routing

I've downgraded from piranha-0.8.4-13.el5.i386 to piranha-0.8.4-11.el5
and everything works as usual
(0010290)
bill_mcgonigle (reporter)
2009-11-05 01:08

confirming on 32-bit. Updating ran the 'service restart' and immediately took down the cluster. This is going to hose anybody running an LVS cluster - can it be pulled from updates until it's fixed?

I don't see an upstream report yet, which is surprising. rpmdiff says:

S.5....T /etc/logrotate.d/piranha-httpd
S.5....T /etc/sysconfig/ha/web/secure/virtual_edit_services.php
S.5....T /usr/sbin/fos
..5....T /usr/sbin/lvsd
S.5....T /usr/sbin/nanny
S.5....T /usr/sbin/pulse
..5....T /usr/sbin/send_arp

That's as far as I've gotten so far - just spent a few hours figuring out why the client's ecommerce was down. :P
(0010291)
bill_mcgonigle (reporter)
2009-11-05 02:42

Comparing the two SRPM's:

Only in piranha-0.8.4-13.el5: piranha-0.8.4-ipvs-rport.patch
Only in piranha-0.8.4-13.el5: piranha-0.8.4-logrotate.patch
Only in piranha-0.8.4-13.el5: piranha-0.8.4-nanny-default.patch
Only in piranha-0.8.4-13.el5: piranha-0.8.4-quotes.patch

The logrotate script should be immaterial, the quotes patch is for PHP. The nanny-default appears to add a default check of an HTTP string, which looks OK at first glance, so I'm focusing on the ipvs-rport patch.

It's tied to Redhat #500909, which "You are not authorized to access bug #500909." Why might that be?
(0010292)
bill_mcgonigle (reporter)
2009-11-05 04:34

filed upstream bug 533113

- Issue History
Date Modified Username Field Change
2009-10-22 13:30 moylo New Issue
2009-10-22 13:30 moylo Assigned To => kbsingh@karan.org
2009-10-22 16:49 moylo Issue Monitored: moylo
2009-10-22 16:50 moylo Note Added: 0010109
2009-11-01 19:29 fabrana Issue Monitored: fabrana
2009-11-01 19:47 fabrana Note Added: 0010230
2009-11-05 01:08 bill_mcgonigle Note Added: 0010290
2009-11-05 02:42 bill_mcgonigle Note Added: 0010291
2009-11-05 04:24 bill_mcgonigle Issue Monitored: bill_mcgonigle
2009-11-05 04:34 bill_mcgonigle Note Added: 0010292


Copyright © 2000 - 2009 Mantis Group
Powered by Mantis Bugtracker