[Pacemaker] Seriously, what's up with the releases?

[Andrew Beekhof]

On Feb 16, 2009, at 2:07 PM, Nikola Ciprich wrote:

> Hi Andrew,
> I've been forced to install latest 1.0 snapshot to one of production  
> servers
> since pingd in 1.0.1 was unusable for me. So I'm already testing  
> what's going
> to be 1.0.2 release :)
> Everything works as expected (even the pingd with multiple NICs).

good to hear :)

>
>
> But today I noticed that there might be one more problem with pingd,  
> it seems
> to be leaking memory somewhere :(
> on my testing machine, pingd uses (I mean RSS, not VIRT) ~160M RAM,  
> and on
> production machine it is already ~360MB, which is quite a lot for  
> small pinging
> daemon :)

indeed.  i've fixed a few memory leaks recently, I'll have to check if  
pingd was affected.

>
> If You have time, could You have a look at it? I'll gladly provide  
> You with
> more information if needed.

i should be able to reproduce locally
you could file a bug though - that would help

>
> with best regards
> nik
>
> PS: I'm replying to this mail for second time, sorry if my mail  
> reached You twice,
> but I'm bit confused about mail address... everybody seems to be  
> using pacemaker at clusterlabs.org,
> (as well as reply address), but mail returned to me with user  
> unknown, and mailmain info says pacemaker at oss.clusterlabs.org
> should be used.

hmmm - odd.  i'll investigate.  thanks!

>
>
> On Fri, Feb 13, 2009 at 06:04:12PM +0100, Andrew Beekhof wrote:
>> Its probably been noticed that I haven't been around much lately, and
>> that the December release is overdue - to put it politely.
>>
>> Rest assured I'm still here working overtime on Pacemaker, but my
>> contractual obligation to get the cluster stack ready for SLES11 are
>> taking up most of my time at the moment.
>>
>> It also hasn't helped that my test cluster kept throwing up new and
>> exciting errors that CTS (not Pacemaker) wasn't equipped to handle -
>> such as syslog logging with the pre-dhcp-assigned hostname :-/
>>
>> Which of course doesn't help those of you hitting bugs in 1.0.1  
>> that I
>> fixed ages ago.
>>
>> So what I propose to do is, now that CTS is running smoothly again,  
>> run
>> the regression tests on Monday/Tuesday and release 1.0.2 with the  
>> caveat
>> that it has only been tested on OpenAIS.  It _should_ continue to  
>> work on
>> Heartbeat, there's no reason why it wouldn't, but I won't have the
>> bandwidth to verify this for the next month or so.
>>
>> Then, with luck, it'll be back to a regular 1.0.3 in March thats been
>> tested with both stacks.
>>
>>
>> Of course if someone wanted to volunteer to do the testing on  
>> Heartbeat,
>> that would also be a significant help :-)
>>
>> Andrew
>>
>> _______________________________________________
>> Pacemaker mailing list
>> Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>
> -- 
> -------------------------------------
> Nikola CIPRICH
> LinuxBox.cz, s.r.o.
> 28. rijna 168, 709 01 Ostrava
>
> tel.:   +420 596 603 142
> fax:    +420 596 621 273
> mobil:  +420 777 093 799
> www.linuxbox.cz
>
> mobil servis: +420 737 238 656
> email servis: servis at linuxbox.cz
> -------------------------------------