[Lustre-discuss] Frequent appearence of LustreError: no handle for file close ino

[Andreas Dilger]

On 2010-04-16, at 01:27, Christos Theodosiou wrote:
> our lustre installation uses two failover MDSes, which serve 10
> file-systems. We recently upgraded from 1.8.1.1 to 1.8.2 version.
>
> By monitoring the MDSes I noticed that we get frequent error messages
> (1-4 times/hour) which looklike this:
>
> Apr 16 10:36:10 lustre01 kernel: LustreError:
> 31386:0:(mds_open.c:1666:mds_close()) @@@ no handle for file close ino
> 4011004: cookie 0x699184f2b7f715d6  req at ffff8100621d2c00
> x1332632126967249/t0 o35->60822f67-b798-f355-d5b1-54d7d7fa9f15@:0/0  
> lens
> 408/472 e 0 to 0 dl 1271403376 ref 1 fl Interpret:/0/0 rc 0/0

This likely means some client was evicted in the past, but later on it  
is closing a file that was previously opened.  This is not an error to  
be concerned about, and we should probably just turn it off.

> o35->60822f67-b798-f355-d5b1-54d7d7fa9f15@:0/0 lens 408/400 e 0 to 0  
> dl

You can find which client this is via (something like, I'm not logged on
my system):

lfs get_param mds.*.exports.*.uuid | grep 60822f67-b798-f355- 
d5b1-54d7d7fa9f15

> 1271403376 ref 1 fl Interpret:/0/0 rc -116/0
> Apr 16 10:36:10 lustre01 kernel: LustreError:
> 31386:0:(ldlm_lib.c:1848:target_send_reply_msg()) Skipped 1 previous
> similar message
>
> and
> Apr 16 11:07:51 lustre02 kernel: LustreError:
> 6305:0:(mds_open.c:1666:mds_close()) @@@ no handle for file close ino
> 2264949: cookie 0xc4e85523602d2686  req at ffff81006787c000
> x1332654888280600/t0 o35->52c78d24-f0da-a56e-9bc5-2e16e6e78790@:0/0  
> lens
> 408/976 e 0 to 0 dl 1271405277 ref 1 fl Interpret:/0/0 rc 0/0
> Apr 16 11:07:51 lustre02 kernel: LustreError:
> 6305:0:(ldlm_lib.c:1848:target_send_reply_msg()) @@@ processing error
> (-116)  req at ffff81006787c000 x1332654888280600/t0
> o35->52c78d24-f0da-a56e-9bc5-2e16e6e78790@:0/0 lens 408/688 e 0 to 0  
> dl
> 1271405277 ref 1 fl Interpret:/0/0 rc -116/0
>
> I would like to ask:
> a) is this a critical error message?
> b) is there any way to find out more info about it. e.g. filesystem,
> filename and lustre client that are related to this error?
> c) is there any way to resolve this errors?
>


Cheers, Andreas
--
Andreas Dilger
Principal Engineer, Lustre Group
Oracle Corporation Canada Inc.