[Gluster-users] high memory usage of mount

Fri Aug 8 07:34:49 UTC 2014

>From the statedump, it looks like the iobufs are not leaking any more.
Inode and dentry have huge hot counts, but it is expected if large number
of files are present and also depends on the kernel parameter 'VFS cache pressure'.
Unable to identify which resource is leaking.

Can you provide the workload(data size, number of files, operations) that is leading to memory leak?
This will help us reproduce and debug.

Regards,
Poornima

----- Original Message -----
From: "Tamas Papp" <tompos at martos.bme.hu>
To: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "Poornima Gurusiddaiah" <pgurusid at redhat.com>
Cc: Gluster-users at gluster.org
Sent: Wednesday, August 6, 2014 5:59:15 PM
Subject: Re: [Gluster-users] high memory usage of mount

Yes, I did.
I have to do it at least once per day.

Currently:

$ free
              total       used       free     shared    buffers cached
Mem:      16422548   16047536     375012          0        320 256884
-/+ buffers/cache:   15790332     632216
Swap:      5859324    3841584    2017740

http://rtfm.co.hu/glusterdump.24405.dump.1407327928.zip

Thanks,
tamas

On 08/06/2014 02:22 PM, Pranith Kumar Karampuri wrote:
> You may have to remount the volume so that the already leaked memory 
> is reclaimed by the system. If you still see the leaks, please provide 
> the updated statedumps.
>
> Pranith
>
> On 08/05/2014 07:30 PM, Tamas Papp wrote:
>> Just an update, the settings below did not help for me.
>>
>> Current settings:
>>
>> Volume Name: w-vol
>> Type: Distribute
>> Volume ID: 89e31546-cc2e-4a27-a448-17befda04726
>> Status: Started
>> Number of Bricks: 5
>> Transport-type: tcp
>> Bricks:
>> Brick1: gl0:/mnt/brick1/export
>> Brick2: gl1:/mnt/brick1/export
>> Brick3: gl2:/mnt/brick1/export
>> Brick4: gl3:/mnt/brick1/export
>> Brick5: gl4:/mnt/brick1/export
>> Options Reconfigured:
>> nfs.mount-udp: on
>> nfs.addr-namelookup: off
>> nfs.ports-insecure: on
>> nfs.port: 2049
>> cluster.stripe-coalesce: on
>> nfs.disable: off
>> performance.flush-behind: on
>> performance.io-thread-count: 64
>> performance.quick-read: off
>> performance.stat-prefetch: on
>> performance.io-cache: off
>> performance.write-behind: on
>> performance.read-ahead: on
>> performance.write-behind-window-size: 4MB
>> performance.cache-refresh-timeout: 1
>> performance.cache-size: 4GB
>> network.frame-timeout: 60
>> performance.cache-max-file-size: 1GB
>>
>>
>> Cheers,
>> tamas
>>
>> On 08/04/2014 09:22 AM, Tamas Papp wrote:
>>> hi Poornima,
>>>
>>> I don't really have any advice, how you could reproduce this issue 
>>> also I don't have coredump (the process killed after oom issue).
>>>
>>> I will see, what can I do.
>>>
>>>
>>> I set the two settings you wrote.
>>>
>>>
>>> Cheers,
>>> tamas
>>>
>>> On 08/04/2014 08:36 AM, Poornima Gurusiddaiah wrote:
>>>> Hi,
>>>>
>>>>  From the statedump it is evident that the iobufs are leaking.
>>>> Also the hot count of the pool-name=w-vol-io-cache:rbthash_entry_t 
>>>> is 10053, implies io-cache xlator could be the cause of the leak.
>>>>  From the logs, it looks like, quick-read performance xlator is 
>>>> calling iobuf_free with NULL pointers, implies quick-read could be 
>>>> leaking iobufs as well.
>>>>
>>>> As a temperory solution, could you disable io-cache and/or 
>>>> quick-read and see if the leak still persists?
>>>>
>>>> $gluster volume set io-cache off
>>>> $gluster volume set quick-read off
>>>>
>>>> This may reduce the performance to certain extent.
>>>>
>>>> For further debugging, could you provide the core dump or steps to 
>>>> reproduce if avaiable?
>>>>
>>>> Regards,
>>>> Poornima
>>>>
>>>> ----- Original Message -----
>>>> From: "Tamas Papp" <tompos at martos.bme.hu>
>>>> To: "Poornima Gurusiddaiah" <pgurusid at redhat.com>
>>>> Cc: Gluster-users at gluster.org
>>>> Sent: Sunday, August 3, 2014 10:33:17 PM
>>>> Subject: Re: [Gluster-users] high memory usage of mount
>>>>
>>>>
>>>> On 07/31/2014 09:17 AM, Tamas Papp wrote:
>>>>> On 07/31/2014 09:02 AM, Poornima Gurusiddaiah wrote:
>>>>>> Hi,
>>>>> hi,
>>>>>
>>>>>> Can you provide the statedump of the process, it can be obtained as
>>>>>> follows:
>>>>>> $ gluster --print-statedumpdir  #create this directory if it doesn't
>>>>>> exist.
>>>>>> $ kill -USR1 <pid-of-glusterfs-process> #generates state dump.
>>>>> http://rtfm.co.hu/glusterdump.2464.dump.1406790562.zip
>>>>>
>>>>>> Also, xporting Gluster via Samba-VFS-plugin method is preferred over
>>>>>> Fuse mount export. For more details refer to:
>>>>>> http://lalatendumohanty.wordpress.com/2014/02/11/using-glusterfs-with-samba-and-samba-vfs-plugin-for-glusterfs-on-fedora-20/ 
>>>>>>
>>>>>>
>>>>> When I tried it about half year ago it didn't work properly. Clients
>>>>> lost mounts, access errors etc.
>>>>>
>>>>> But I will give it a try, though it's not included in ubuntu's samba
>>>>> AFAIK.
>>>>>
>>>>>
>>>>> Thank you,
>>>>> tamas
>>>>>
>>>>> ps. I forget to mention, I can see this issue only one node. The rest
>>>>> of nodes are fine.
>>>> hi Poornima,
>>>>
>>>> Do you have  idea, what's going on here?
>>>>
>>>> Thanks,
>>>> tamas
>>>
>>> _______________________________________________
>>> Gluster-users mailing list
>>> Gluster-users at gluster.org
>>> http://supercolony.gluster.org/mailman/listinfo/gluster-users
>>
>> _______________________________________________
>> Gluster-users mailing list
>> Gluster-users at gluster.org
>> http://supercolony.gluster.org/mailman/listinfo/gluster-users
>