root / hspace.1 @ 5d4f9eed
History | View | Annotate | Download (12.5 kB)
1 |
.TH HSPACE 1 2009-06-01 htools "Ganeti H-tools" |
---|---|
2 |
.SH NAME |
3 |
hspace \- Cluster space analyzer for Ganeti |
4 |
|
5 |
.SH SYNOPSIS |
6 |
.B hspace |
7 |
.B "[backend options...]" |
8 |
.B "[algorithm options...]" |
9 |
.B "[request options..."] |
10 |
.B "[-p]" |
11 |
.B "[-v... | -q]" |
12 |
|
13 |
.B hspace |
14 |
.B --version |
15 |
|
16 |
.TP |
17 |
Backend options: |
18 |
.BI " -m " cluster |
19 |
| |
20 |
.BI " -L[" path "]" |
21 |
| |
22 |
.BI " -n " nodes-file |
23 |
.BI " -i " instances-file |
24 |
| |
25 |
.BI " --simulate " spec |
26 |
|
27 |
.TP |
28 |
Algorithm options: |
29 |
.BI "[ --max-cpu " cpu-ratio " ]" |
30 |
.BI "[ --min-disk " disk-ratio " ]" |
31 |
.BI "[ -O " name... " ]" |
32 |
|
33 |
.TP |
34 |
Request options: |
35 |
.BI "[--memory " mem "]" |
36 |
.BI "[--disk " disk "]" |
37 |
.BI "[--req-nodes " req-nodes "]" |
38 |
.BI "[--vcpus " vcpus "]" |
39 |
.BI "[--tiered-alloc " spec "]" |
40 |
|
41 |
|
42 |
.SH DESCRIPTION |
43 |
hspace computes how many additional instances can be fit on a cluster, |
44 |
while maintaining N+1 status. |
45 |
|
46 |
The program will try to place instances, all of the same size, on the |
47 |
cluster, until the point where we don't have any N+1 possible |
48 |
allocation. It uses the exact same allocation algorithm as the hail |
49 |
iallocator plugin. |
50 |
|
51 |
The output of the program is designed to interpreted as a shell |
52 |
fragment (or parsed as a \fIkey=value\fR file). Options which extend |
53 |
the output (e.g. -p, -v) will output the additional information on |
54 |
stderr (such that the stdout is still parseable). |
55 |
|
56 |
The following keys are available in the output of the script (all |
57 |
prefixed with \fIHTS_\fR): |
58 |
.TP |
59 |
.I SPEC_MEM, SPEC_DSK, SPEC_CPU, SPEC_RQN |
60 |
These represent the specifications of the instance model used for |
61 |
allocation (the memory, disk, cpu, requested nodes). |
62 |
|
63 |
.TP |
64 |
.I CLUSTER_MEM, CLUSTER_DSK, CLUSTER_CPU, CLUSTER_NODES |
65 |
These represent the total memory, disk, CPU count and total nodes in |
66 |
the cluster. |
67 |
|
68 |
.TP |
69 |
.I INI_SCORE, FIN_SCORE |
70 |
These are the initial (current) and final cluster score (see the hbal |
71 |
man page for details about the scoring algorithm). |
72 |
|
73 |
.TP |
74 |
.I INI_INST_CNT, FIN_INST_CNT |
75 |
The initial and final instance count. |
76 |
|
77 |
.TP |
78 |
.I INI_MEM_FREE, FIN_MEM_FREE |
79 |
The initial and final total free memory in the cluster (but this |
80 |
doesn't necessarily mean available for use). |
81 |
|
82 |
.TP |
83 |
.I INI_MEM_AVAIL, FIN_MEM_AVAIL |
84 |
The initial and final total available memory for allocation in the |
85 |
cluster. If allocating redundant instances, new instances could |
86 |
increase the reserved memory so it doesn't necessarily mean the |
87 |
entirety of this memory can be used for new instance allocations. |
88 |
|
89 |
.TP |
90 |
.I INI_MEM_RESVD, FIN_MEM_RESVD |
91 |
The initial and final reserved memory (for redundancy/N+1 purposes). |
92 |
|
93 |
.TP |
94 |
.I INI_MEM_INST, FIN_MEM_INST |
95 |
The initial and final memory used for instances (actual runtime used |
96 |
RAM). |
97 |
|
98 |
.TP |
99 |
.I INI_MEM_OVERHEAD, FIN_MEM_OVERHEAD |
100 |
The initial and final memory overhead - memory used for the node |
101 |
itself and unacounted memory (e.g. due to hypervisor overhead). |
102 |
|
103 |
.TP |
104 |
.I INI_MEM_EFF, HTS_INI_MEM_EFF |
105 |
The initial and final memory efficiency, represented as instance |
106 |
memory divided by total memory. |
107 |
|
108 |
.TP |
109 |
.I INI_DSK_FREE, INI_DSK_AVAIL, INI_DSK_RESVD, INI_DSK_INST, INI_DSK_EFF |
110 |
Initial disk stats, similar to the memory ones. |
111 |
|
112 |
.TP |
113 |
.I FIN_DSK_FREE, FIN_DSK_AVAIL, FIN_DSK_RESVD, FIN_DSK_INST, FIN_DSK_EFF |
114 |
Final disk stats, similar to the memory ones. |
115 |
|
116 |
.TP |
117 |
.I INI_CPU_INST, FIN_CPU_INST |
118 |
Initial and final number of virtual CPUs used by instances. |
119 |
|
120 |
.TP |
121 |
.I INI_CPU_EFF, FIN_CPU_EFF |
122 |
The initial and final CPU efficiency, represented as the count of |
123 |
virtual instance CPUs divided by the total physical CPU count. |
124 |
|
125 |
.TP |
126 |
.I INI_MNODE_MEM_AVAIL, FIN_MNODE_MEM_AVAIL |
127 |
The initial and final maximum per-node available memory. This is not |
128 |
very useful as a metric but can give an impression of the status of |
129 |
the nodes; as an example, this value restricts the maximum instance |
130 |
size that can be still created on the cluster. |
131 |
|
132 |
.TP |
133 |
.I INI_MNODE_DSK_AVAIL, FIN_MNODE_DSK_AVAIL |
134 |
Like the above but for disk. |
135 |
|
136 |
.TP |
137 |
.I TSPEC |
138 |
If the tiered allocation mode has been enabled, this parameter holds |
139 |
the pairs of specifications and counts of instances that can be |
140 |
created in this mode. The value of the key is a space-separated list |
141 |
of values; each value is of the form \fImemory,disk,vcpu=count\fR |
142 |
where the memory, disk and vcpu are the values for the current spec, |
143 |
and count is how many instances of this spec can be created. A |
144 |
complete value for this variable could be: \fB4096,102400,2=225 |
145 |
2560,102400,2=20 512,102400,2=21\fR. |
146 |
|
147 |
.TP |
148 |
.I ALLOC_USAGE |
149 |
The current usage represented as initial number of instances divided |
150 |
per final number of instances. |
151 |
|
152 |
.TP |
153 |
.I ALLOC_COUNT |
154 |
The number of instances allocated (delta between FIN_INST_CNT and |
155 |
INI_INST_CNT). |
156 |
|
157 |
.TP |
158 |
.I ALLOC_FAIL*_CNT |
159 |
For the last attemp at allocations (which would have increased |
160 |
FIN_INST_CNT with one, if it had succeeded), this is the count of the |
161 |
failure reasons per failure type; currently defined are FAILMEM, |
162 |
FAILDISK and FAILCPU which represent errors due to not enough memory, |
163 |
disk and CPUs, and FAILN1 which represents a non N+1 compliant cluster |
164 |
on which we can't allocate instances at all. |
165 |
|
166 |
.TP |
167 |
.I ALLOC_FAIL_REASON |
168 |
The reason for most of the failures, being one of the above FAIL* |
169 |
strings. |
170 |
|
171 |
.TP |
172 |
.I OK |
173 |
A marker representing the successful end of the computation, and |
174 |
having value "1". If this key is not present in the output it means |
175 |
that the computation failed and any values present should not be |
176 |
relied upon. |
177 |
|
178 |
.PP |
179 |
|
180 |
If the tiered allocation mode is enabled, then many of the INI_/FIN_ |
181 |
metrics will be also displayed with a TRL_ prefix, and denote the |
182 |
cluster status at the end of the tiered allocation run. |
183 |
|
184 |
.SH OPTIONS |
185 |
The options that can be passed to the program are as follows: |
186 |
|
187 |
.TP |
188 |
.BI "--memory " mem |
189 |
The memory size of the instances to be placed (defaults to 4GiB). |
190 |
|
191 |
.TP |
192 |
.BI "--disk " disk |
193 |
The disk size of the instances to be placed (defaults to 100GiB). |
194 |
|
195 |
.TP |
196 |
.BI "--req-nodes " num-nodes |
197 |
The number of nodes for the instances; the default of two means |
198 |
mirrored instances, while passing one means plain type instances. |
199 |
|
200 |
.TP |
201 |
.BI "--vcpus " vcpus |
202 |
The number of VCPUs of the instances to be placed (defaults to 1). |
203 |
|
204 |
.TP |
205 |
.BI "--max-cpu " cpu-ratio |
206 |
The maximum virtual-to-physical cpu ratio, as a floating point number |
207 |
between zero and one. For example, specifying \fIcpu-ratio\fR as |
208 |
\fB2.5\fR means that, for a 4-cpu machine, a maximum of 10 virtual |
209 |
cpus should be allowed to be in use for primary instances. A value of |
210 |
one doesn't make sense though, as that means no disk space can be used |
211 |
on it. |
212 |
|
213 |
.TP |
214 |
.BI "--min-disk " disk-ratio |
215 |
The minimum amount of free disk space remaining, as a floating point |
216 |
number. For example, specifying \fIdisk-ratio\fR as \fB0.25\fR means |
217 |
that at least one quarter of disk space should be left free on nodes. |
218 |
|
219 |
.TP |
220 |
.B -p, --print-nodes |
221 |
Prints the before and after node status, in a format designed to allow |
222 |
the user to understand the node's most important parameters. |
223 |
|
224 |
The node list will contain these informations: |
225 |
.RS |
226 |
.TP |
227 |
.B F |
228 |
a character denoting the status of the node, with '-' meaning an |
229 |
offline node, '*' meaning N+1 failure and blank meaning a good node |
230 |
.TP |
231 |
.B Name |
232 |
the node name |
233 |
.TP |
234 |
.B t_mem |
235 |
the total node memory |
236 |
.TP |
237 |
.B n_mem |
238 |
the memory used by the node itself |
239 |
.TP |
240 |
.B i_mem |
241 |
the memory used by instances |
242 |
.TP |
243 |
.B x_mem |
244 |
amount memory which seems to be in use but cannot be determined why or |
245 |
by which instance; usually this means that the hypervisor has some |
246 |
overhead or that there are other reporting errors |
247 |
.TP |
248 |
.B f_mem |
249 |
the free node memory |
250 |
.TP |
251 |
.B r_mem |
252 |
the reserved node memory, which is the amount of free memory needed |
253 |
for N+1 compliance |
254 |
.TP |
255 |
.B t_dsk |
256 |
total disk |
257 |
.TP |
258 |
.B f_dsk |
259 |
free disk |
260 |
.TP |
261 |
.B pcpu |
262 |
the number of physical cpus on the node |
263 |
.TP |
264 |
.B vcpu |
265 |
the number of virtual cpus allocated to primary instances |
266 |
.TP |
267 |
.B pri |
268 |
number of primary instances |
269 |
.TP |
270 |
.B sec |
271 |
number of secondary instances |
272 |
.TP |
273 |
.B p_fmem |
274 |
percent of free memory |
275 |
.TP |
276 |
.B p_fdsk |
277 |
percent of free disk |
278 |
.TP |
279 |
.B r_cpu |
280 |
ratio of virtual to physical cpus |
281 |
.TP |
282 |
.B lCpu |
283 |
the dynamic CPU load (if the information is available) |
284 |
.TP |
285 |
.B lMem |
286 |
the dynamic memory load (if the information is available) |
287 |
.TP |
288 |
.B lDsk |
289 |
the dynamic disk load (if the information is available) |
290 |
.TP |
291 |
.B lNet |
292 |
the dynamic net load (if the information is available) |
293 |
.RE |
294 |
|
295 |
.TP |
296 |
.BI "-O " name |
297 |
This option (which can be given multiple times) will mark nodes as |
298 |
being \fIoffline\fR, and instances won't be placed on these nodes. |
299 |
|
300 |
Note that hspace will also mark as offline any nodes which are |
301 |
reported by RAPI as such, or that have "?" in file-based input in any |
302 |
numeric fields. |
303 |
.RE |
304 |
|
305 |
.TP |
306 |
.BI "-n" nodefile ", --nodes=" nodefile |
307 |
The name of the file holding node information (if not collecting via |
308 |
RAPI), instead of the default \fInodes\fR file (but see below how to |
309 |
customize the default value via the environment). |
310 |
|
311 |
.TP |
312 |
.BI "-i" instancefile ", --instances=" instancefile |
313 |
The name of the file holding instance information (if not collecting |
314 |
via RAPI), instead of the default \fIinstances\fR file (but see below |
315 |
how to customize the default value via the environment). |
316 |
|
317 |
.TP |
318 |
.BI "-m" cluster |
319 |
Collect data not from files but directly from the |
320 |
.I cluster |
321 |
given as an argument via RAPI. If the argument doesn't contain a colon |
322 |
(:), then it is converted into a fully-built URL via prepending |
323 |
https:// and appending the default RAPI port, otherwise it's |
324 |
considered a fully-specified URL and is used as-is. |
325 |
|
326 |
.TP |
327 |
.BI "-L[" path "]" |
328 |
Collect data not from files but directly from the master daemon, which |
329 |
is to be contacted via the luxi (an internal Ganeti protocol). An |
330 |
optional \fIpath\fR argument is interpreted as the path to the unix |
331 |
socket on which the master daemon listens; otherwise, the default path |
332 |
used by ganeti when installed with "--localstatedir=/var" is used. |
333 |
|
334 |
.TP |
335 |
.BI "--simulate " description |
336 |
Instead of using actual data, build an empty cluster given a node |
337 |
description. The \fIdescription\fR parameter must be a comma-separated |
338 |
list of four elements, describing in order: |
339 |
|
340 |
.RS |
341 |
|
342 |
.RS |
343 |
.TP |
344 |
the number of nodes in the cluster |
345 |
|
346 |
.TP |
347 |
the disk size of the nodes, in mebibytes |
348 |
|
349 |
.TP |
350 |
the memory size of the nodes, in mebibytes |
351 |
|
352 |
.TP |
353 |
the cpu core count for the nodes |
354 |
|
355 |
.RE |
356 |
|
357 |
An example description would be \fB20,102400,16384,4\fR describing a |
358 |
20-node cluster where each node has 100GiB of disk space, 16GiB of |
359 |
memory and 4 CPU cores. Note that all nodes must have the same specs |
360 |
currently. |
361 |
|
362 |
.RE |
363 |
|
364 |
.TP |
365 |
.BI "--tiered-alloc " spec |
366 |
Beside the standard, fixed-size allocation, also do a tiered |
367 |
allocation scheme where the algorithm starts from the given |
368 |
specification and allocates until there is no more space; then it |
369 |
decreases the specification and tries the allocation again. The |
370 |
decrease is done on the matric that last failed during allocation. The |
371 |
specification given is similar to the "--simulate" option and it |
372 |
holds: |
373 |
|
374 |
.RS |
375 |
|
376 |
.RS |
377 |
|
378 |
.TP |
379 |
the disk size of the instance |
380 |
|
381 |
.TP |
382 |
the memory size of the instance |
383 |
|
384 |
.TP |
385 |
the vcpu count for the insance |
386 |
|
387 |
.RE |
388 |
|
389 |
An example description would be \fB10240,8192,2\fR describing an |
390 |
initial starting specification of 10GiB of disk space, 4GiB of memory |
391 |
and 2 VCPUs. |
392 |
|
393 |
Also note that the normal allocation and the tiered allocation are |
394 |
independent, and both start from the initial cluster state; as such, |
395 |
the instance count for these two modes are not related one to another. |
396 |
|
397 |
.RE |
398 |
|
399 |
.TP |
400 |
.B -v, --verbose |
401 |
Increase the output verbosity. Each usage of this option will increase |
402 |
the verbosity (currently more than 2 doesn't make sense) from the |
403 |
default of one. At verbosity 2 the location of the new instances is |
404 |
shown in the standard error. |
405 |
|
406 |
.TP |
407 |
.B -q, --quiet |
408 |
Decrease the output verbosity. Each usage of this option will decrease |
409 |
the verbosity (less than zero doesn't make sense) from the default of |
410 |
one. |
411 |
|
412 |
.TP |
413 |
.B -V, --version |
414 |
Just show the program version and exit. |
415 |
|
416 |
.SH EXIT STATUS |
417 |
|
418 |
The exist status of the command will be zero, unless for some reason |
419 |
the algorithm fatally failed (e.g. wrong node or instance data). |
420 |
|
421 |
.SH BUGS |
422 |
|
423 |
The algorithm is highly dependent on the number of nodes; its runtime |
424 |
grows exponentially with this number, and as such is impractical for |
425 |
really big clusters. |
426 |
|
427 |
The algorithm doesn't rebalance the cluster or try to get the optimal |
428 |
fit; it just allocates in the best place for the current step, without |
429 |
taking into consideration the impact on future placements. |
430 |
|
431 |
.SH ENVIRONMENT |
432 |
|
433 |
If the variables \fBHTOOLS_NODES\fR and \fBHTOOLS_INSTANCES\fR are |
434 |
present in the environment, they will override the default names for |
435 |
the nodes and instances files. These will have of course no effect |
436 |
when the RAPI or Luxi backends are used. |
437 |
|
438 |
.SH SEE ALSO |
439 |
.BR hbal "(1), " hscan "(1), " ganeti "(7), " gnt-instance "(8), " |
440 |
.BR gnt-node "(8)" |
441 |
|
442 |
.SH "COPYRIGHT" |
443 |
.PP |
444 |
Copyright (C) 2009 Google Inc. Permission is granted to copy, |
445 |
distribute and/or modify under the terms of the GNU General Public |
446 |
License as published by the Free Software Foundation; either version 2 |
447 |
of the License, or (at your option) any later version. |
448 |
.PP |
449 |
On Debian systems, the complete text of the GNU General Public License |
450 |
can be found in /usr/share/common-licenses/GPL. |