| Option |
Default |
Description |
|
[ClusterManager]
|
| service |
tashi.clustermanager.ClusteManagerService |
This specifies the class name of the cluster manager service |
| data |
tashi.clustermanager.data.GetentOverride |
This specifies the class name for the data backend. Available options are: GetentOverride, LdapOverride, Pickled, FromConfig, and SQL (all in tashi.clustermanager.data) |
| dfs |
tashi.dfs.Vfs |
This specifies the class name of the DFS implementation used at the cluster manager |
| publisher |
tashi.messaging.GangliaPublisher |
This specifies the class name of the publisher used at the cluster manager |
| nodeManagerPort |
9883 |
This specifies what port number to try to connect to Node Managers on |
|
[ClusterManagerService]
|
| convertExceptions |
True |
Whether exceptions raised in ClusterManagerService should be converted to generic exceptions and passed along to the client |
| port |
9882 |
What port to listen on for client requests |
| expireHostTime |
30.0 |
The amount of time (in seconds) to let pass before considering a host down if the cluster manager has not heard from it |
| allowDecayed |
30.0 |
The amount of time to let an instance or a host be decayed before attempting to sync |
| allowMismatchedVersions |
False |
Whether to allow Node Managers that have different software versions to join the cluster |
| maxMemory |
7680 |
The maximum amount of memory any user can request for a VM |
| maxCores |
8 |
The maximum number of cores any user can request for a VM |
| allowDuplicateNames |
False |
Whether VMs are allowed to have duplicate names |
|
[Accounting]
|
| publisher |
tashi.messaging.GangliaPublisher |
This specifies the class name of the publisher used at the accounting server |
|
[GetentOverride]
|
| baseData |
tashi.clustermanager.data.Pickled |
The backend data class to use for data not related to users |
| fetchThreshold |
60.0 |
The minimum amount of time to wait between calling "getent" |
|
[LdapOverride]
|
| baseData |
tashi.clustermanager.data.Pickled |
The backend data class to use for data not related to users |
| fetchThreshold |
3600.0 |
The minimum amount of time to wait between issuing LDAP queries |
| nameKey |
sAMAccountName |
The variable name that represents usernames in the LDAP results |
| idKey |
msSFU30UidNumber |
The variable name that represents user ids in the LDAP results |
| ldapCommand |
ldapsearch -x -w password -h host -b searchbase -D binddn msSFU30LoginShell=* -z 0 |
The ldap command to run to perform lookups |
|
[FromConfig]
|
| hostn |
Host(d={'id':1,'name':'blade043'}) |
Specifies a host object to be created |
| networkn |
Network(d={'id':1,'name':'global'}) |
Specifies a network object to be created |
| usern |
User(d={'id':1,'name':'mryan3'}) |
Specifies a user object to be created |
|
[Pickled]
|
| file |
/var/tmp/cm.dat |
The file to use to store pickled data for the cluster manager |
|
[SQL]
|
| uri |
mysql://root@localhost/tashi |
The URI that describes what SQL backend to use -- examples include sqlite://filename and mysql://user@host/dbname |
| password |
changeme |
The password to use when connecting to a MySQL database |
|
[NodeManager]
|
| dfs |
tashi.dfs.Vfs |
This specifies the class name of the DFS implementation used at the node manager |
| vmm |
tashi.nodemanager.vmcontrol.Qemu |
This specifies the class name of the VMM backend used at the node manager |
| service |
tashi.nodemanager.NodeManagerService |
This specifies the class name of the node manager service |
| publisher |
tashi.messaging.GangliaPublisher |
This specifies the class name of the publisher used at the node manager |
|
[NodeManagerService]
|
| convertExceptions |
True |
Whether exceptions raised in NodeManagerService should be converted to generic exceptions and passed along to the client |
| port |
9883 |
What port to listen on for client requests |
| registerHost |
False |
Should the nodemanager auto-register the host with the cluster manager? |
| registerFrequency |
10.0 |
How many seconds should the node manager wait between registering with the cluster manager? |
| clusterManagerHost |
localhost |
What host to connect to in order to find the cluster manager |
| clusterManagerPort |
9882 |
What port to connect to the cluster manager on |
| statsInterval |
0 |
How often should the node manager output per-VM statistics using the publisher |
|
[Qemu]
|
| qemuBin |
/usr/bin/kvm |
The location of the Qemu/Kqemu/KVM binary to be used |
| infoDir |
/var/tmp/VmControlQemu/ |
A location into which the Qemu backend places per-VM control information |
| statsInterval |
0 |
How often should per-VM statistics be collected |
| scratchDir |
/tmp |
Location where Qemu snapshot disk images can be placed |
| scratchVg |
None |
Volume group where host-local logical volumes can be allocated |
| reservedMem |
512 |
Memory to reserve for the host (in megabytes) |
| pollDelay |
1 |
How often to check and see if a VM has exited |
| migrationRetries |
10 |
How many retries are allowed on failed migrations |
| monitorTimeout |
60 |
How long to wait before timing out on monitor command execution |
| migrateTimeout |
300 |
How long to wait before timing out a live migration |
| maxParallelMigrations |
10 |
How many parallel live migrations are allowed |
| useMigrateArgument |
False |
(DEPRECATED) An option to allow the use of "migrate=True" as an argument to KVM -- requires a patch to KVM |
| suspendHandler |
gzip |
Machine state being saved will be piped through this program |
| resumeHandler |
zcat |
Machine state being loaded will be piped through this program |
|
[XenPV]
|
| vmNamePrefix |
tashi |
The name prefix to use when dealing with Xen domains? |
| transientdir |
/tmp |
A place to put transient deltas of disk images |
| defaultVmType |
kernel |
This specifies the type and kernel loading mechanism for the VM -- options are "pvgrub", "pygrub", "kernel", and "hvm" |
| defaultKernel |
/boot/vmlinuz-xen |
If the VmType is "kernel", this is the kernel to load for use in the guest |
| defaultRamdisk |
/boot/initrd-xen |
If the VmType is "kernel", this is the initrd to load for use in the guest |
| defaultDiskType |
qcow |
Default disk image format, e.g. qcow, vhd |
| defaultBridgeFormat |
br%s |
Format string to combine with a network ID to obtain a bridge name |
|
[Vfs]
|
| prefix |
/dfs |
A location into which all DFS access is directed -- this is usually a mount point to common storage |
|
[LocalityService]
|
| host |
localityserver |
The hostname clients use to find the locality server |
| port |
9884 |
The port to listen on for requests |
| staticLayout |
/location/of/layout/file |
The location of a layout specification file |
|
[Client]
|
| clusterManagerHost |
localhost |
The hostname to use to find the cluster manager |
| clusterManagerPort |
9882 |
The port to use when communicating with the cluster manager |
| clusterManagerTimeout |
5.0 |
The number of seconds to wait before assuming an operation has timed out |
|
[Primitive]
|
| hookn |
tashi.agents.DhcpDns |
This allows the specification of an ordered list of agent hooks that perform actions when VMs are started and stopped |
| scheduleDelay |
2.0 |
How long to delay between scheduling cycles |
| densePack |
False |
Should VMs be densely packed across hosts |
| publisher |
tashi.messaging.GangliaPublisher |
The publisher to use for the scheduling agent |
|
[DhcpDns]
|
| dnsKeyFile |
/location/of/private/key/for/dns |
The location of the private key file for DNS updates |
| dnsServer |
1.2.3.4 53 |
The server to connect to to perform DNS updates |
| dnsDomain |
your.domain.com |
The domain in which VM hostnames are inserted |
| dnsExpire |
31536000 |
How long are entries cacheable |
| dhcpServer |
1.2.3.4 |
The server with to which DHCP updates are sent |
| dhcpKeyName |
OMAPI |
The name of the DHCP key being used |
| dhcpSecretKey |
ABcdEf12GhIJKLmnOpQrsT== |
The DHCP secret key |
| ipRangen |
172.16.128.2-172.16.255.254 |
For network n, this specifies the IP range to be used for allocations |
| reverseDns |
True |
Should reverse DNS entries be inserted and removed |
|
[GangliaPublisher]
|
| dmax |
60 |
The amount of time that published messages are held on to before discarding |
| retry |
3600 |
How often, in seconds, to retry executing gmetric to see if it has been installed |
