-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathnode_search.1
161 lines (136 loc) · 7.63 KB
/
node_search.1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
.\" Manpage for node_search.py.
.\" Contact [email protected] to correct typos or errors.
.TH man 1 "9 JULY 2018" "1.2.2" "node_search man page"
.SH NAME
node_search \- get node, user, or host-group information from Univa Grid Engine
.SH SYNOPSIS
\fBnode_search.sh \fR[\fB-h\fR, \fB--help\fR] [\fB-d\fR, \fB--debug\fR] [\fB-a\fR, \fB--all\fR]
[\fB-g\fR, \fB--general_access\fR] [\fB-H\fR, \fB --hosts\fR] [\fB-hostname\fR]
[\fB-u\fR, \fB--user username\fR] [\fB-uf username\fR] [\fB-U\fR] [\fB-v\fR, \fB--visual\fR]
.SH DESCRIPTION
\fRnode_search gathers information from the Univa Grid Engine and displays that information to
stdout, which can then be redirected as desired. The following information can be obtained:
." Next Paragraph
.PP
\fRQueue/Host-group core usage, User Job information, which Host-groups a User belongs to, which
Host-groups exist, which users are recognized by UGE currently.
." BEGIN OPTIONS !!
.SH OPTIONS
.TP
\fB-h\fR, \fB--help
\fRShow a small help dialogue with a short description of each option and a few examples.
." Next Option . . .
.TP
\fB-d\fR, \fB--debug
\fRDisplay information pertaining to the Debug queue such as core usage (free cores, disabled cores,
used cores) and node usage (free nodes, disabled nodes, used nodes). This option allows the optional
\fB--details \fRflag or it can take the \fB-v\fR/\fB--visual \fRflag.
." Next Option . . .
.TP
\fB-g\fR, \fB--general_access
\fRDisplay information pertaining to the General Access hostgroup such as core usage (free cores,
disabled cores, used cores) and node usage (free nodes, disabled nodes, used nodes). This option
allows the optional \fB--details \fRflag or it can take the \fB-v\fR/\fB--visual\fR flag. The general
access hostgroup makes up the core of the long queue, every user can submit jobs to the general access
hostgroup. Keep in mind that while submitting jobs to the long queue (\fB#$ -q long\fR) faculty owned
hostgroups which a user has access to will be targeted first. To find which hostgroups a user has access
to see \fB-uf\fR.
." Next Option . . .
.TP
\fB-H\fR, \fB--hosts
\fRShows all known hosts configured to be used by the Univa Grid Engine. \fINot all of the host-groups
may be available to you, see \fB-uf
." Next Option . . .
.TP
\fB-[hostname]
\fRThis option takes in a variable hostname, which can be found by using \fB-H\fR, and gives the
same information found by using \fB-d \fRor \fB-g \fRi.e. core usage and node usage information
can be found and displayed from that specified hostname. Even if you cannot submit jobs to this
specific host-group, you can still view this information. Don't get confused. \fBNOTE:\fR You
do not need the \fB"@"\fR while specifying the hostname. (i.e. -corke not -@corke)
." Next Option . . .
.TP
\fB-a\fR,\fB --all
\fRDisplay usage information on all nodes detected by UGE. This searches every host-group with a
running total of usage, while discarding any duplicate machines. Duplicate machines occur when
there are nested hostgroups. The \fB--details\fR option can be used with this as well.
." Next Option . . .
.TP
\fB-u\fR, \fB--user \fR[\fBusername\fR]
\fRDisplay job information along with what node and core information pertaining to the specifed
user after \fB-u \fRor \fB--user\fR. The maximum Memory used detected by the grid engine will also
be shown. This option will take the optional flag \fB-details\fR for a more detailed output. The
detailed output will display all of the jobs running on the nodes that the specified user has jobs
running on. Since our nodes are a shared resource, it may be useful to see this if a job is running
poorly. \fBNOTE:\fR jobs running on GPUs will not have the GPU memory accessed.
." Next Option . . .
.TP
\fB-uf \fR[\fBusername\fR]
\fRShow which host-groups and user-lists the specified user after \fB-uf \fRbelongs to. The
information displayed will indicate which host-groups the user has access to, i.e. which host-groups
that user can submit jobs to using the UGE. The user-list aspect of this option is not entirely
useful, but those user-lists are what is used in conifguring qconf to tell the UGE which users belong
to which host-groups.
." Next Option . . .
.TP
\fB-U
\fRDifferent from \fB-u\fR, \fB-U \fRwill display all users currently interacting with the UGE. This
means those users: have jobs running somewhere controlled by the UGE, are waiting in a queue controlled
by the UGE, or recently interacted with the UGE in some way. Thus, not every user who has access to
the UGE will be displayed from this. Do not fret if your name is not on this list.
.TP
\fB-qlong
\fRThis unary flag will parse all nodes which the current user found by the environment's \fB$USER\fR
has access to through UGE. All nodes within hostgroups of the long queue which the user has access to will
be displayed including the total core count, the used core count, free core count, disabled core count,
total nodes, used nodes, free nodes, and error/disabled node count.
.SH OPTIONAL FLAGS
.TP
\fB-v\fR, \fB--visual
\fRDisplay a visual representation of a queue. In this visual representation, \fB[0]\fR represents
an open core, \fB[#]\fR represents a disabled core, and \fB[~]\fR represents a used or taken core.
This flag can be passed in after using \fBnodeSearch -d, -g, \fRor \fB-[hostname]\fR. See
Examples below.
." Next Option . . .
.TP
\fB--details
\fRShow a more detailed output of whatever was passed into \fRnode_search\fR. The result of
\fB--details\fR depends on what it is modifying. If it is modifying the debug queue, long queue, or a
specifed host-group, then the output will show all of the machines/nodes that belong to that queue
or host-group. Each one of those nodes will also show its core usage (used vs total). This will be
the same behavior that occurs if \fB--details\fR is used to modify \fB-a \fR or\fB --all\fR. If \fB--details
\fRis modifying a user, then all of the nodes the specified user is running on will be displayed along with all
of the processes that user owns and the respective memory usage, CPU%, and run time of each process. This is
useful for when a job submitted through UGE spawns or starts multiple processes which are not necessarily tracked
by UGE. If the user has any pending jobs then those will be shown as well. If \fB--details \fRis modifying the
\fB-uf \fRoption, then it will display all of the nodes which belong to the host-groups the specified user has access to.
." END OPTIONS !!
.SH EXAMPLES
.TP
\fBnodeSearch -g
\fRDisplay information pertaining to the long queue (general access host-group only)
.TP
\fBnodeSearch --debug --details
\fRDisplay detailed information pertaining to the debug queue, this will yield information such as
the name of all of the nodes currently occupying the debug queue and their core information.
.TP
\fBnodeSearch -general_access -v
\fRDisplay a visual representation of the host-group '@general_access'. Notice how the '@' is not
needed for the host-group.
.TP
\fBnodeSearch -uf JohnDoe33 --details
\fRDisplay a detailed view of the host-groups that user JohnDoe33 belongs to. This will show each
and every node that also belongs to those host-groups. It may be wise to pipe this output through
something along the lines of less/more. e.g. nodeSearch -uf JohnDoe33 --details | less
.TP
\fBnodeSearch -u JohnDoe
\fRDisplay the user JohnDoe and their jobs along with core usage and the nodes the jobs are running
on. Will display total memory usage per node as used by user JohnDoe.
.TP
\fBnodeSearch -crc_d12chas
\fRDisplay usage statistics for the hostgroup @crc_d12chas.
.SH BUGS
\fRNo known bugs at the current moment in time.
Report any new findings on github: https://github.com/CodyKank/node_search
.SH AUTHOR
\fRCody Kankel ([email protected])