Path /store and /config not working with cifs mount on docker host about docker-recorder

[zutewas]

Update with other test which works.

user@virtubuntu:/etc/otrecorder$ docker run -it --rm -v /media/nas/docker/otr:/mnt busybox
/ # cd /mnt/
/mnt # ls -l
total 0
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 config
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 store
/mnt # mkdir test
/mnt # ls -l
total 0
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 config
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 store
drwxr-xr-x    2 1000     1000             0 Dec 16 09:41 test
/mnt # rm -R test/
/mnt # ls -l
total 0
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 config
drwxr-xr-x    2 1000     1000             0 Dec 16 09:20 store
/mnt #

[jpmens]

A quick search leaves me thinking few people use CIFS with docker and those who do don't have an easy time with it...

Have you tried having the host mount the CIFS FS before launching docker? This might help a bit.

(I have zero experience with CIFS under docker.)

[zutewas]

Thanks for the first hint. I'm using autofs and this mounts on demand. I'm using round about 10 dockers with micro services for private use and there is at the moment no other docker image with this type of problem. I will check your link and play with different settings. ;-)

[jpmens]

Something ocurred to me early this morning: the Recorder uses LMDB which in turn memory-maps (nmap) files. It is absolutely possible that memory mapping files through docker and CIFS is just not possible and the reason why some component crashes and we dump core.

I wouldn't expect it is us because we handle an error but maybe some other layer:

rc = mdb_env_create(&gc->env);
if (rc != 0) {
        olog(LOG_ERR, "gcache_open: mdb_env_create: %s", mdb_strerror(rc))
;
        free(gc);
        return (NULL);
}

mdb_env_set_mapsize(gc->env, lmdb_size);

[jpmens]

Closing this as stale.