Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-13 Thread Andrew Beekhof
On Thu, Aug 13, 2009 at 11:25 AM, Lars Marowsky-Bree wrote: > On 2009-08-11T15:44:33, Andrew Beekhof wrote: > >> at the pacemaker level there is the dc-deadtime option though, so the >> new node will at least wait a while before trying to shoot anyone. > > Yeah, but dc-deadtime is permanent, and n

Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-13 Thread Lars Marowsky-Bree
On 2009-08-11T15:44:33, Andrew Beekhof wrote: > at the pacemaker level there is the dc-deadtime option though, so the > new node will at least wait a while before trying to shoot anyone. Yeah, but dc-deadtime is permanent, and not just for the first formation. Regards, Lars -- Architect

Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-12 Thread Tim Serong
Lars Marowsky-Bree wrote: > On 2009-08-11T00:41:42, Tim Serong wrote: > > Hi Tim, > > it's worth noting that we hunted down an additional new scenario which > could cause this to trigger. > > Some switches seem to take noticably longer to get multicast > communication up, while unicast or even

Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-11 Thread Andrew Beekhof
On Tue, Aug 11, 2009 at 3:39 PM, Lars Marowsky-Bree wrote: > On 2009-08-11T00:41:42, Tim Serong wrote: > > Hi Tim, > > it's worth noting that we hunted down an additional new scenario which > could cause this to trigger. > > Some switches seem to take noticably longer to get multicast > communicat

Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-11 Thread Lars Marowsky-Bree
On 2009-08-11T00:41:42, Tim Serong wrote: Hi Tim, it's worth noting that we hunted down an additional new scenario which could cause this to trigger. Some switches seem to take noticably longer to get multicast communication up, while unicast or even broadcast are already operational. So despit

Re: [Pacemaker] STONITH Deathmatch Explained

2009-08-10 Thread Tim Serong
I wrote: >>> I've written up a brief document entitled "STONITH Deathmatch Explained >>> (and Some Hints for Resource Agent Authors and Systems Engineers)": >>> >>> http://ourobengr.com/ha >>> >>> ... Then Dejan Muhamedagic wrote: >> ... >> >> - in "Causes ..." you missed to mention split-brain

Re: [Pacemaker] STONITH Deathmatch Explained

2009-05-14 Thread Andreas Mock
> -Ursprüngliche Nachricht- > Von: "Joe Armstrong" > Gesendet: 14.05.09 18:01:11 > An: "'pacema...@clusterlabs.org'" > Betreff: Re: [Pacemaker] STONITH Deathmatch Explained > I agree - a nice read. You might want to also add a pos

Re: [Pacemaker] STONITH Deathmatch Explained

2009-05-14 Thread Joe Armstrong
>On Thu, May 14, 2009 at 06:32:00PM +1000, Tim Serong wrote: >> Greetings, >> >> I've written up a brief document entitled "STONITH Deathmatch Explained >> (and Some Hints for Resource Agent Authors and Systems Engineers)": >> >> http://ourobengr.com/ha >> >> It's a description of causes of ST

Re: [Pacemaker] STONITH Deathmatch Explained

2009-05-14 Thread Dejan Muhamedagic
Hi, On Thu, May 14, 2009 at 06:32:00PM +1000, Tim Serong wrote: > Greetings, > > I've written up a brief document entitled "STONITH Deathmatch Explained > (and Some Hints for Resource Agent Authors and Systems Engineers)": > > http://ourobengr.com/ha > > It's a description of causes of STONIT

[Pacemaker] STONITH Deathmatch Explained

2009-05-14 Thread Tim Serong
Greetings, I've written up a brief document entitled "STONITH Deathmatch Explained (and Some Hints for Resource Agent Authors and Systems Engineers)": http://ourobengr.com/ha It's a description of causes of STONITH deathmatch in Heartbeat/Pacemaker HA clusters, where two nodes continually shoo