[Samba] Index Corruption xBase database

[Ray Klassen]

from the man page. I've never tried veto oplocks. But it looks like 
something that might be useful...

>        veto oplock files (S)
> 
>            This parameter is only valid when the oplocks parameter is turned
>            on for a share. It allows the Samba administrator to selectively
>            turn off the granting of oplocks on selected files that match a
>            wildcarded list, similar to the wildcarded list used in the veto
>            files parameter.
> 
>            You might want to do this on files that you know will be heavily
>            contended for by clients. A good example of this is in the NetBench
>            SMB benchmark program, which causes heavy client contention for
>            files ending in .SEM. To cause Samba not to grant oplocks on these
>            files you would use the line (either in the [global] section or in
>            the section for the particular NetBench share.
> 
>            An example of use is:
> 
>                veto oplock files = /.*SEM/
> 
>            Default: veto oplock files =  # No files are vetoed for oplock
>            grants


On 2018-11-22 10:49 a.m., Jose Concha via samba wrote:
> Thanks for the tip. Is veto oplocks different to oplocks = no?
> 
> Just now I run into an error, but this time I have this log:
> 
> ../lib/util/util.c:401(fcntl_lock)
> fcntl_lock: fcntl lock gave errno 11 (resource temporally unavailable)
> fcntl_lock: lock failed at offset 2147483646 count 1 op 37 type 1 (resource
> temporally unavailable)
> 
> Does this give you any clue of how the problem can be solved?
> 
> The offset is the number I see on smbstatus.
> 
> Thanks for your help
> 
> El jue., 22 nov. 2018 a las 15:28, Ray Klassen (<
> julius_ahenobarbus at yahoo.co.uk>) escribió:
> 
>> Still pursuing the idea of oplocks, the 'veto oplocks =' parameter might
>> be your friend. It seems that you can specify exact file names disable
>> oplocking on just those files...
>>
>>
>> On 2018-11-22 9:35 a.m., Jose Concha via samba wrote:
>>> Using oplocks = no makes the communication extremely slow. For testing we
>>> have set level2 oplocks = no, but it haven't made much difference.
>>>
>>> Any clue about the smbstatus -B output?
>>>
>>> El jue., 22 nov. 2018 a las 14:00, Ray Klassen via samba (<
>>> samba at lists.samba.org>) escribió:
>>>
>>>> Are the shares set to oplocks = no ? That sometimes has a role in
>>>> database corruption.
>>>>
>>>> On 2018-11-22 6:31 a.m., Jose Concha via samba wrote:
>>>>> Hello,
>>>>>
>>>>> I've been working with a xBase database (similar to dBase) in a
>> multiuser
>>>>> application. There are like 40 users working on the database on the
>> same
>>>>> time reading and like 10 writing information.
>>>>>
>>>>> Sometimes the CDX files (Index Files) get corrupted and is it's
>> necessary
>>>>> to pack the tables (rebuild the indexes).
>>>>>
>>>>> I have notice using the smbstatus -B command the following
>>>>>
>>>>> Pid       dev:inode          R/W      start                  size
>>>> name
>>>>> 4429    80a:b60b34:0      W      1073746530        1         PARAMS.DBF
>>>>> 3627    80a:b60b1d:0     PW     2147483646        1        MOVCTA.cdx
>>>>> 3445    80a:b60b1d:0     PW     2147483646        1        MOVCTA.cdx
>>>>> 6078    80a:b60b1d:0     PW     2147483646        1        MOVCTA.cdx
>>>>> 4429    80a:b60b1d:0     PW     2147483646        1        MOVCTA.cdx
>>>>>
>>>>> Everytime PW appears and the start number 2147483646 indexes get
>>>> corrupted.
>>>>> I've been searching documentation about smbstatus -B without success,
>> so
>>>> I
>>>>> do not know what start means or what can be causing the problem.
>>>>>
>>>>> I am running in a Giga ethernet LAN with Cat 6. I am using version
>> 4.5.12
>>>>> on Debian 9
>>>>>
>>>>> Thanks for your help
>>>>>
>>>>
>>>>
>>>> --
>>>> To unsubscribe from this list go to the following URL and read the
>>>> instructions:  https://lists.samba.org/mailman/options/samba
>>
>>