1 # EC2 configuration for Arvados Node Manager.
2 # All times are in seconds unless specified otherwise.
5 # The dispatcher can customize the start and stop procedure for
6 # cloud nodes. For example, the SLURM dispatcher drains nodes
7 # through SLURM before shutting them down.
10 # Node Manager will ensure that there are at least this many nodes running at
11 # all times. If node manager needs to start new idle nodes for the purpose of
12 # satisfying min_nodes, it will use the cheapest node type. However, depending
13 # on usage patterns, it may also satisfy min_nodes by keeping alive some
14 # more-expensive nodes
17 # Node Manager will not start any compute nodes when at least this
21 # Upper limit on rate of spending (in $/hr), will not boot additional nodes
22 # if total price of already running nodes meets or exceeds this threshold.
23 # default 0 means no limit.
26 # Poll EC2 nodes and Arvados for new information every N seconds.
29 # Polls have exponential backoff when services fail to respond.
30 # This is the longest time to wait between polls.
33 # If Node Manager can't succesfully poll a service for this long,
34 # it will never start or stop compute nodes, on the assumption that its
35 # information is too outdated.
36 poll_stale_after = 600
38 # If Node Manager boots a cloud node, and it does not pair with an Arvados
39 # node before this long, assume that there was a cloud bootstrap failure and
40 # shut it down. Note that normal shutdown windows apply (see the Cloud
41 # section), so this should be shorter than the first shutdown window value.
42 boot_fail_after = 1800
44 # "Node stale time" affects two related behaviors.
45 # 1. If a compute node has been running for at least this long, but it
46 # isn't paired with an Arvados node, do not shut it down, but leave it alone.
47 # This prevents the node manager from shutting down a node that might
48 # actually be doing work, but is having temporary trouble contacting the
50 # 2. When the Node Manager starts a new compute node, it will try to reuse
51 # an Arvados node that hasn't been updated for this long.
52 node_stale_after = 14400
54 # File path for Certificate Authorities
55 certs_file = /etc/ssl/certs/ca-certificates.crt
59 file = /var/log/arvados/node-manager.log
61 # Log level for most Node Manager messages.
62 # Choose one of DEBUG, INFO, WARNING, ERROR, or CRITICAL.
63 # WARNING lets you know when polling a service fails.
64 # INFO additionally lets you know when a compute node is started or stopped.
67 # You can also set different log levels for specific libraries.
68 # Pykka is the Node Manager's actor library.
69 # Setting this to DEBUG will display tracebacks for uncaught
70 # exceptions in the actors, but it's also very chatty.
73 # Setting apiclient to INFO will log the URL of every Arvados API request.
77 host = zyxwv.arvadosapi.com
81 # Accept an untrusted SSL certificate from the API server?
87 # It's usually most cost-effective to shut down compute nodes during narrow
88 # windows of time. For example, EC2 bills each node by the hour, so the best
89 # time to shut down a node is right before a new hour of uptime starts.
90 # Shutdown windows define these periods of time. These are windows in
91 # full minutes, separated by commas. Counting from the time the node is
92 # booted, the node WILL NOT shut down for N1 minutes; then it MAY shut down
93 # for N2 minutes; then it WILL NOT shut down for N3 minutes; and so on.
94 # For example, "54, 5, 1" means the node may shut down from the 54th to the
95 # 59th minute of each hour of uptime.
96 # Specify at least two windows. You can add as many as you need beyond that.
97 shutdown_windows = 54, 5, 1
106 # This section defines filters that find compute nodes.
107 # Tags that you specify here will automatically be added to nodes you create.
108 # Replace colons in Amazon filters with underscores
109 # (e.g., write "tag:mytag" as "tag_mytag").
110 instance-state-name = running
111 tag_arvados-class = dynamic-compute
115 # New compute nodes will send pings to Arvados at this host.
116 # You may specify a port, and use brackets to disambiguate IPv6 addresses.
117 ping_host = hostname:port
119 # Give the name of an SSH key on AWS...
122 # ... or a file path for an SSH key that can log in to the compute node.
123 # (One or the other, not both.)
126 # The EC2 IDs of the image and subnet compute nodes should use.
130 # Comma-separated EC2 IDs for the security group(s) assigned to each
132 security_groups = idstring1, idstring2
135 # You can define any number of Size sections to list EC2 sizes you're
136 # willing to use. The Node Manager should boot the cheapest size(s) that
137 # can run jobs in the queue.
139 # Each size section MUST define the number of cores are available in this
140 # size class (since libcloud does not provide any consistent API for exposing
142 # You may also want to define the amount of scratch space (expressed
143 # in GB) for Crunch jobs. You can also override Amazon's provided
144 # data fields (such as price per hour) by setting them here.