[Glass] shrpcmonitor fails to start because lock file doesn't exits

[Dale Henrichs]

On Thu, Jan 23, 2014 at 8:14 AM, Mariano Martinez Peck <
marianopeck at gmail.com> wrote:

>
>
>
> On Thu, Jan 23, 2014 at 12:50 PM, Dale Henrichs <
> dale.henrichs at gemtalksystems.com> wrote:
>
>> Mariano,
>>
>> I think these lines should be diagnostic:
>>
>>   |   GemStone could not retrieve the IPC identifier associated with the
>> memory |
>>   |   key -704639710.  shmget() error = errno=28,ENOSPC, There is no
>> space left |
>>   | on the device (or, in fcntl(), there are no more record locks).
>>       |
>>
>> Well hidden, but I think it is telling you that you don't have enough
>> shared memory on the system where you are starting your second stone ... if
>> you have plenty of RAM on the system you probably need to bump up the size
>> of shared memory
>>
>>
> Dale, I feel a bit stupid...I saw that, but I ignored it thinking it was
> impossible. Also, I found some logs from gemstone that tells me something
> is wrong but it actually isn't so I wasn't very trustful ;)  I can point
> you to these places once I find them again if you want.
>
> Anyway....I don't understand why that error. My virtualmachine (virtualbox
> machine) has 8GB assigned. I read in the admin guide that assiging a 0.75
> of such for the shared memory is a good idea. So, 6GB is approx. that. So I
> have:
>
> $ sudo sysctl -A | grep kernel.shm
> kernel.shmmax = 6442450944
> kernel.shmall = 1572864
> kernel.shmmni = 4096
> kernel.shm_rmid_forced = 0
>
> Which from what I understand, it is more than enough right?  Also:
>
> $ free -m
>              total       used       free     shared    buffers     cached
> Mem:          7870       4679       3191          0        163       2561
> -/+ buffers/cache:       1954       5916
> Swap:        12223          0      12223
>
> so I have 3GB free of RAM and 12 GB of swap space.
>
>
> $ sudo ipcs -m
> 0x93000d26 163845     stone1 660        2156756992 15
> 0xb000130c 196614     stone2 660        2156756992 0
>
> Here it seems both stones are as assigned the 2GB shared memory.
>
> so...what do I miss?
>
I'm not exactly sure what I am seeing here ... you show that there are two
shared memory segments: stone1 and stone2. Now are these 2 running stones
 and the failure occurs when you try to add the third?

Sometimes shared memory gets allocated and if you kill the cache process,
the shared memory does not get correctly detached and shared memory stays
allocated on the system ... it is safe to kill the stoned process but the
shrpcmonitor is not safe to kill ...

>
> And now I have also more questions...then thing is like this: I may have
> several stones running. I want to assign to them the max possible SPC, that
> is, 2GB. Are those 2GB took from the beginning or on demand? Say I have 6GB
> of RAM...does it mean I can have (almost) 3 stones only? Or I can have many
> ones and if I have less RAM, then even if the SPC is 2GB max..they will be
> smaller...?
>

with 6GB of shared memory allocated you can only have ~3 2GB SPC running
... I say approximately because  because GemStone may actually allocate a
bit more than 2GB of shared memory when you ask for a 2GB shared page cache
... the shared memory is completely allocated ... so you need to manually
manage your SPC sizes

Dale
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gemtalksystems.com/mailman/private/glass/attachments/20140124/5dd7b412/attachment.html>