Torque 6 install from source

Compile node requires:

# yum -y groupinstall "development tools"
# yum -y install boost-devel gcc-c++ hwloc \
hwloc-devel libxml2-devel openssl-devel \
pam-devel readline-devel tk-devel

Configuration flags:

# ./configure --enable-maxdefault --enable-cgroups \
--with-loadlibfile --disable-gui

–disable-gui require for Torque 6 else build process fails…

Build:

# make && make packages

Install:
Supporting packages required on pbs_server / pbs_mom / lazy:

# yum -y install hwloc-libs tcl
# yum -y install libcgroup-tools
# yum -y install hwloc-libs libcgroup-tools tcl

Then –install packages to your nodes.
(Note: On node where you plan to run torque.setup (usually pbs_server node or login / client node), please make sure client package is also installed — required for trqauth.)

Configuration for lib installed at default location:

# echo '/usr/local/lib' > /etc/ld.so.conf.d/torque.conf && ldconfig

If you did not configure and build packages for this site, you will need to edit $PBS_HOME/server_name on all nodes

First time pbs_server startup:

# $torque_source/torque.setup root && qterm

Make sure you qterm otherwise you won’t be able to startup using systemd normally.

Copy $PBS_HOME/pbs_environment from server to MOMs
Copy the required systemd unit file from $torque_source/contrib/systemd; start all required components using systemd.

Extra:

# qmgr -c 'set server auto_node_np = True'

For automatically setting NP in nodes file.

libvirtd: cannot start existing lvm pool

libvirtd: unsupported configuration: cannot find any matching source devices for logical volume group 'dummy_vg'
libvirtd: internal error: Failed to autostart storage pool 'dummy_vg': unsupported configuration: cannot find any matching source devices for logical volume group 'dummy_vg'

Workaround from: https://bugzilla.redhat.com/show_bug.cgi?id=1411600

But besides “device path", I also have to remove capacity, allocation and available before the pool would start.

qemu-img create image on GlusterFS via libgfapi

# qemu-img create -f qcow2 -b /vm-readonly/centos7-minimal.qcow2 \
gluster://10.181.0.1/vm-image/fleetx-703.qcow2

Back image access is done via FUSE since it’s on read-only volume and Gluster-FUSE client may easily cache the relatively small base image shared among all VMs. This also preserve the flexibility for moving the back image directory to other POSIX compatible file systems in case such needs arise in the future; since we want to avoid relocating back image as it is a trivial but fragile procedure.

Adding OST to Lustre

1. Add fstab entry and OST mount point

/dev/vdb /lustre-ost02 lustre _netdev 0 0

2. Confirm currently configured Lustre parameters

# tunefs.lustre --dryrun /dev/vda

3. Format OST with correct index, fsname, & mgs (with dryrun)

# mkfs.lustre --ost --index=2 --fsname=hpclustre --mgsnode=10.0.0.1@o2ib --dryrun /dev/vdb

4. Mount new OST

# mount /dev/vdb

Logstash mutate-join does not run when mutate-merge is used

Please note the internal execution order of mutate plugin:

    rename(event) if @rename
    update(event) if @update
    replace(event) if @replace
    convert(event) if @convert
    gsub(event) if @gsub
    uppercase(event) if @uppercase
    lowercase(event) if @lowercase
    strip(event) if @strip
    remove(event) if @remove
    split(event) if @split
    join(event) if @join
    merge(event) if @merge

    filter_matched(event)

Reference

Work around is to use another instance of mutate plugin…

libvirt: failed to connect to gluster://… No such file or directory

Encountered on CentOS 7 / GlusterFS 3.8 / libvirt-daemon 2.0, with the following work around:

Step 1: Add the following option to /etc/glusterfs/glusterd.vol on all gluster servers

    option rpc-auth-allow-insecure on

Step 2: glusterd must be restarted in order to activate the option

# systemctl restart glusterd

Step 3: Add the following option to the concerned gluster volume

# gluster volume set gv0 server.allow-insecure on

This option is required since GlusterFS 3.7.1

If you are still encountering issue, please make sure you are running latest libvirt-daemon (libvirt-daemon-driver-storage).