[CTDB] A question about reloading CTDB

[曹宇]

Thanks Ronnie

Well, yes it's true that the scenario is rare.
But since we are building a project, so we may need to answer the
question that how long will it finish.
Because users may not have enough patience.

Is there some other way reaching the goal?
e.g.
Change some configuration to shorten the time?


于 2011-6-20 16:27, ronnie sahlberg 写道:
> That command may need to wait across a number of recoveries and each
> recovery will have a grace period of 10 seconds before the next
> recovery can happen. So with sascading recoveries it can well take
> multiples of 10 seconds to complete.
>
>
> Since you only use this command when you add/remove physical nodes
> from a cluster,
> which would be something that only happens exceptionally seldom, one a year?,
> why does it matter if it takes a while to run ?
>
> regards
> ronnie sahlberg
>
> 2011/6/20 曹宇 <caoyu at huaweisymantec.com>:
>> Hi Samba team,
>>
>>   We have updated CTDB to version 1.0.114-2 (should be the latest version),
>>   and noticed that with 1.0.114-2, the command "ctdb reloadnodes"
>>   executes for longer than 10sec as well.
>>
>>   Does anyone have some ideas for solving this problem?
>>
>>
>> At 2011-6-10 11:09, Cao Yu Wrote:
>>> Hi Samba team
>>>
>>> I'm using CTDB version 1.0.114-1 for a samba cluster.
>>> We first set up a CTDB node, then tried to add a new CTDB node
>>> with the "ctdb reloadnodes" command.
>>>
>>> But I found out that sometimes, the command executes for a fairly long
>>> time(longer than 10 sec).
>>>
>>> Could any one tell me is it possible that the "ctdb reloadnodes" command
>>> finish within 8 sec?
>>> Plus, when will the command fail?
>>>
>>> Thanks
>>>
>>> Cao Yu
>>