git-yww commented 2 years ago

We created a new pod with a lvm pvc, it started successfully and everything seemed ok.

With 'lvs' command executed in the csi container, we could find the new lv: bash-5.0# lvs | grep vg01 pvc-dcd29be5-41c6-46b0-bd81-93e45056ee95 vg01 -wi-ao---- 4.00g However, when we executed 'lvs' command on host, no lv existed... [ivan@k8s-node16 ~]$ sudo lvs | grep vg01 [ivan@k8s-node16 ~]$

Also，the vg we used showed differences: vgs in csi container: bash-5.0# vgs | grep vg01 vg01 1 1 0 wz--n- <19.53g <15.53g vgs on host: [ivan@k8s-node16 ~]$ sudo vgs | grep vg01 vg01 1 0 0 wz--n- <19.53g <19.53g

But, on the host, we could find the new lv under the /dev/vg01 and /dev/mapper. [ivan@k8s-node16 ~]$ sudo ls /dev/vg01 pvc-dcd29be5-41c6-46b0-bd81-93e45056ee95 [ivan@k8s-node16 ~]$ sudo ls /dev/mapper | grep vg01 vg01-pvc--dcd29be5--41c6--46b0--bd81--93e45056ee95

Is this related to the diff of lvm version in container and on host, or something else like the centos7 os of host? lvm version in container: bash-5.0# lvm version LVM version: 2.02.186(2) (2019-08-27) Library version: 1.02.164 (2019-08-27) Driver version: 4.39.0 Configuration: ./configure --build=x86_64-alpine-linux-musl --host=x86_64-alpine-linux-musl --prefix=/usr --sysconfdir=/etc --libdir=/lib --sbindir=/sbin --localstatedir=/var --disable-nls --disable-readline --enable-pkgconfig --enable-applib --with-thin=internal --enable-dmeventd --enable-cmdlib --with-thin-check=/sbin/thin_check --with-thin-dump=/sbin/thin_dump --with-thin-repair=/sbin/thin_repair --with-dmeventd-path=/sbin/dmeventd --enable-udev_rules CLDFLAGS=-Wl,--as-needed lvm version on host: LVM version: 2.02.171(2)-RHEL7 (2017-05-03) Library version: 1.02.140-RHEL7 (2017-05-03) Driver version: 4.39.0 Configuration: ./configure --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --with-default-dm-run-dir=/run --with-default-run-dir=/run/lvm --with-default-pid-dir=/run --with-default-locking-dir=/run/lock/lvm --with-usrlibdir=/usr/lib64 --enable-lvm1_fallback --enable-fsadm --with-pool=internal --enable-write_install --with-user= --with-group= --with-device-uid=0 --with-device-gid=6 --with-device-mode=0660 --enable-pkgconfig --enable-applib --enable-cmdlib --enable-dmeventd --enable-blkid_wiping --enable-python2-bindings --with-cluster=internal --with-clvmd=corosync --enable-cmirrord --with-udevdir=/usr/lib/udev/rules.d --enable-udev_sync --with-thin=internal --enable-lvmetad --with-cache=internal --enable-lvmpolld --enable-lvmlockd-dlm --enable-lvmlockd-sanlock --enable-dmfilemapd

It would be greatly appreciated if anyone could help.

git-yww commented 2 years ago

173 Found this related issue with a "pvscan --cache" solution. However, every time we provision a new lv, pvscan --cache command will need to be executed once on the host machine. Obviously, that's not an elegant solution.

git-yww commented 2 years ago

Propose another way here:

Before execute the lvcreate command, we first enter the mount namespace of the host machine by nsenter. To see host mount namespace in the lvm node container, a "hostPID=true" is also needed for the lvm node pod spec. After some tests, it appeared that the problem of lv invisibility could be well solved by this solution.

abhilashshetty04 commented 11 months ago

Hi @git-yww , Thanks for reporting . However, we see no difference in lvs and vgs output in container and host on lvm driver v1.3.0 and Ubuntu 20.0.4. If it is reproducible on 1.3 as well let us know.

dsharma-dc commented 3 months ago

From the information given, the lvm version is 2.02 and has --enable-lvmetad which is enabling caching of metadata. Hence unless pvscan is done, the info wasn't refreshed. With lvm driver 2.03, this caching daemon behaviour is removed, and metadata is read form the disk instead of cache. Please confirm if that resolves the issue, or if still using 2.02 please disable the daemon and try again.

Meanwhile I'm closing this issue, please feel free to update any new information or reopen in case of any issues around this.

openebs / lvm-localpv

LVs are invisible on the host machine #193

173 Found this related issue with a "pvscan --cache" solution. However, every time we provision a new lv, pvscan --cache command will need to be executed once on the host machine. Obviously, that's not an elegant solution.