[PATCH] LMDB full patch set
Stefan Metzmacher
metze at samba.org
Tue May 15 08:17:05 UTC 2018
Am 15.05.2018 um 10:06 schrieb Stefan Metzmacher via samba-technical:
> Am 15.05.2018 um 06:50 schrieb Andrew Bartlett via samba-technical:
>> On Mon, 2018-05-14 at 08:02 +0200, Stefan Metzmacher via samba-
>> technical wrote:
>>> Am 14.05.2018 um 06:59 schrieb Andrew Bartlett via samba-technical:
>>>> On Fri, 2018-05-11 at 20:07 +0200, Stefan Metzmacher wrote:
>>>>> If it would be optional for sometime and it would actually work
>>>>> it would be much easier to agree to push it to master.
>>>>
>>>> I've made it optional. Builds of ldb and Samba have a --without-ldb-
>>>> lmdb option. Builds --without-ad-dc do not enable lmdb, and only AD DC
>>>> sefltest builds require it.
>>>>
>>>> Otherwise, the only other restriction is that 'make test' on ldb is
>>>> incompatible with --without-ldb-lmdb. This pre-empts the 'tests gone
>>>> missing' issue. Both the python and the cmocka tests require lmdb, so
>>>> it is clearer to just require it.
>>>>
>>>> The ad_dc and vampire_dc environments in Samba's selftest now use
>>>> --backend-store=mdb
>>>>
>>>> autobuild confirms that ldb will compile --without-ldb-lmdb
>>>>
>>>> If an AD DC build is made against a system ldb without lmdb support,
>>>> some tests will fail.
>>>>
>>>> https://gitlab.com/catalyst-samba/samba/pipelines/21959419
>>>> https://gitlab.com/catalyst-samba/samba/commits/metze-master4-lmdb-full
>>>>
>>>> Finally, an ldb make test passes on all the ldb_mdb commits.
>>>>
>>>> Please review and push ;-)
>>>
>>> Thanks! I'll have a look later.
>>
>> I've reworked the patches to hook this up to the join etc, and the new
>> pipeline is here:
>>
>> https://gitlab.com/catalyst-samba/samba/pipelines/22016682
>>
>>> In the meantime can you please try a private autobuild on sn-devel-144?
>>
>> I ran one, it may be due to the number of other autobuilds being run by
>> the same user as the ldb tests passed previously but failed now.
>>
>> (So while they all failed, some failed with a different flapping test,
>> well past the ldb tests).
>>
>> Are there unusual limits on mmap() areas on that host?
>>
>> In the meantime I've updated the error message and dropped the size to
>> 8GB for now. Having this be a relatively fixed number just for sn-
>> devel is not ideal however.
>
> I guess we need to remove the ulimit -v.
>
> On my system top shows this:
>
> 29037 metze 39 19 80,238g 109276 38468 S 0,3 0,7 0:22.95 samba
Here's a more complete list:
29052 metze 39 19 80,259g 102180 31780 S 0,0 0,6 0:01.12 smbd
29064 metze 39 19 0,157t 98896 18696 S 0,0 0,6 0:00.21
winbindd
28282 metze 39 19 0,157t 98232 18484 S 0,0 0,6 0:00.06
winbindd
29066 metze 39 19 0,157t 98016 17828 S 0,0 0,6 0:00.04
winbindd
28195 metze 39 19 0,157t 97808 17432 S 0,0 0,6 0:00.33
samba
28283 metze 39 19 0,157t 97740 18004 S 0,0 0,6 0:00.00
winbindd
28190 metze 39 19 0,157t 96156 15784 S 0,0 0,6 0:00.09
samba
29043 metze 39 19 80,218g 95448 25600 S 0,0 0,6 0:00.56
winbindd
29067 metze 39 19 0,157t 95428 14728 S 0,0 0,6 0:00.10 lpqd
28199 metze 39 19 0,157t 94676 14320 S 0,0 0,6 0:00.05
samba
28208 metze 39 19 80,217g 94628 25228 S 0,0 0,6 0:00.36
winbindd
28201 metze 39 19 0,157t 94112 13752 S 0,0 0,6 0:00.00
samba
29065 metze 39 19 0,157t 93752 13576 S 0,0 0,6 0:00.08
winbindd
28186 metze 39 19 80,226g 93744 23700 S 0,0 0,6 0:00.22
samba
28284 metze 39 19 0,157t 93052 13340 S 0,0 0,6 0:00.03
winbindd
28191 metze 39 19 0,157t 91832 11476 S 0,0 0,6 0:00.02
samba
28194 metze 39 19 0,157t 91148 10792 S 0,0 0,6 0:00.00 samba
metze
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20180515/64b3583d/signature.sig>
More information about the samba-technical
mailing list