[Samba] problems after lowering tombstone...

Achim Gottinger achim at ag-web.biz
Thu Jul 24 02:37:06 MDT 2014


Am 23.07.2014 21:50, schrieb Achim Gottinger:
> Am 23.07.2014 21:28, schrieb mourik jan heupink - merit:
>> Hi Achim, list,
>>
>> Interesting, yes, exactly that seems the case. So...hence my advise 
>> against brtfs.
>>
>> Stefan told me that perhaps we could lower tombstone in 10-day steps, 
>> and wait after each step until everything settles down.
> This is an good advice! Took around 25 mins to delete 1000 entries 
> here. Had 72000 deleted entries here with 180 days so aprx. 60000 are 
> older than 30 days. Means 25 hrs.
> Looking forward to an bunch of issues and phone calls tomorrow.
>>
>> Now I only hope for some good answers for my (just posted) sssd 
>> question, and then hopefully some quieter time ahead...
> Can't help you with that. I'm on debian wheezy shipped sssd is too old 
> tried to backport it was not too difficult but too much work to do it 
> on an regular basis.
>
>
> achim~
>>
>> Mourik Jan
>>
>>> You know the odd thing is i just decreased tomstone lifetime from 180
>>> down to 30 days on an setup with windbind issues in hope to get some
>>> speed improvements.
>>>
>>> The samba process is still busy ~90% and if i run samba-toor drs
>>> showrepl i get the same error while still having 5GB free on /var.
>>>
>>> I assume your db corruption was caused by no space on /var (due to 
>>> brtfs
>>> issues).
>>>
>>> ==== INBOUND NEIGHBORS ====
>>>
>>> ERROR(runtime): DsReplicaGetInfo of type 0 failed - (-1073610723,
>>> 'NT_STATUS_RPC_PROTOCOL_ERROR')
>>>    File "/usr/lib/python2.7/dist-packages/samba/netcmd/drs.py", line
>>> 116, in drsuapi_ReplicaInfo
>>>      (info_type, info) =
>>> ctx.drsuapi.DsReplicaGetInfo(ctx.drsuapi_handle, 1, req1)
>>>
>
Had increased the tombstone lifetime back to 170 days and after all 
deleted object where purged the error was gone. Continued decreasing 
lifetime in 10-20 day steps.
Usualy samba need a few minutes (~5) before the purgin starts after the 
lifetime value was decreased. An restart of samba triggers the purge 
immediate.




More information about the samba mailing list