[Linux-ha-dev] RFE - Don't consider missing client a
fatal error(?)
Alan Robertson
alanr at unix.sh
Tue Oct 12 16:22:16 MDT 2004
Scott Herod wrote:
> On Mon, 2004-10-11 at 18:09, Lars Ellenberg wrote:
>
>>/ 2004-10-11 16:52:40 -0600
>>\ Scott Herod:
>>
>>>Hello all,
>>>
>>>I've got a client that I load with a respawn command in my ha.cf file.
>>>Currently (heartbeat-1.2.3) the heartbeat refuses to start if the client
>>>binary is missing. Is there away to tell the heartbeat to consider a
>>>missing client only worthy of a warning? If not, could such a thing be
>>>added?
>>
>>if the client you configured is not available, that is fatal.
>>if you had not wanted it HA, you had not configured it.
>>if you know it is not available,
>>you can comment it out/remove it from the config.
>>when you later add it, you can tell heartbeat to "reload" the config.
>>
>>or did I miss the point?
>>
>> lge
>
>
> Yep. That about sums it up. :-(
>
> I confess that this is all for a system that is to be automatically
> deployed in which case the client had better be there. The problem I'm
> actually running into is other developers who whine about having to
> install the client ... Okay, I withdraw the request and will refer other
> whiners to method 5 of protecting syslogd from DoS attacks.
I don't see that this is a huge deal one way or the other, but (since I
wrote this code), I lean towards the current solution.
Heartbeat will eventually stop respawning it if it's respawning too fast
without stopping heartbeat itself. This situation is somewhat analagous to
the "missing client" situation.
On the other hand, the missing client business is pretty gross by
comparison - and might just involve a misspelling error - and refusing to
start is bound to get someone's attention so they will fix it - whereas
just logging a message is much less likely to get someone's attention to
fix it.
--
Alan Robertson <alanr at unix.sh>
"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce
More information about the Linux-HA-Dev
mailing list