summaryrefslogtreecommitdiffstats
path: root/scheduler/nufa/src/nufa.c
blob: 92b70ca758086023707fbe38a525cf9c6dcd0ba2 (plain)
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
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
/*
  Copyright (c) 2006-2009 Gluster, Inc. <http://www.gluster.com>
  This file is part of GlusterFS.

  GlusterFS is free software; you can redistribute it and/or modify
  it under the terms of the GNU General Public License as published
  by the Free Software Foundation; either version 3 of the License,
  or (at your option) any later version.

  GlusterFS is distributed in the hope that it will be useful, but
  WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  General Public License for more details.

  You should have received a copy of the GNU General Public License
  along with this program.  If not, see
  <http://www.gnu.org/licenses/>.
*/

#ifndef _CONFIG_H
#define _CONFIG_H
#include "config.h"
#endif

#include <sys/time.h>

#include "scheduler.h"
#include "common-utils.h"

struct nufa_sched_struct {
	xlator_t *xl;
	struct timeval last_stat_fetch;
	int64_t free_disk;
	int32_t refresh_interval;
	unsigned char eligible;
};

struct nufa_struct {
	struct nufa_sched_struct *array;
	struct timeval last_stat_fetch;

	int32_t *local_array; /* Used to keep the index of the local xlators */
	int32_t local_xl_index; /* index in the above array */
	int32_t local_xl_count; /* Count of the local subvolumes */

	uint32_t refresh_interval;
	uint32_t min_free_disk;
  
	gf_lock_t nufa_lock;
	int32_t child_count;
	int32_t sched_index;  
};

#define NUFA_LIMITS_MIN_FREE_DISK_DEFAULT    15
#define NUFA_REFRESH_INTERVAL_DEFAULT        30

static int32_t
nufa_init (xlator_t *xl)
{
	int32_t index = 0;
	data_t *local_name = NULL;
	data_t *data = NULL;
	xlator_list_t *trav_xl = xl->children;
	struct nufa_struct *nufa_buf = NULL;

	nufa_buf = CALLOC (1, sizeof (struct nufa_struct));
	ERR_ABORT (nufa_buf);

	data = dict_get (xl->options, "scheduler.limits.min-free-disk");
	if (data) {
		if (gf_string2percent (data->data, 
				       &nufa_buf->min_free_disk) != 0) {
			gf_log ("nufa", GF_LOG_ERROR, 
				"invalid number format \"%s\" of "
				"\"option scheduler.limits.min-free-disk\"", 
				data->data);
			return -1;
		}
		if (nufa_buf->min_free_disk >= 100) {
			gf_log ("nufa", GF_LOG_ERROR,
				"check \"option scheduler.limits.min-free-disk"
				"\", it should be percentage value");
			return -1;
		}
	} else {
		gf_log ("nufa", GF_LOG_WARNING, 
			"No option for limit min-free-disk given, "
			"defaulting it to 15%%");
		nufa_buf->min_free_disk = NUFA_LIMITS_MIN_FREE_DISK_DEFAULT;
	}
	data = dict_get (xl->options, "scheduler.refresh-interval");
	if (data != NULL) {
                if (gf_string2time (data->data, 
				    &nufa_buf->refresh_interval) != 0) {
                        gf_log ("nufa", GF_LOG_ERROR, 
                                "invalid number format \"%s\" of "
                                "\"option scheduler.refresh-interval\"", 
                                data->data);
                        return -1;
                }
        } else {
		gf_log ("nufa", GF_LOG_WARNING, 
			"No option for scheduler.refresh-interval given, "
			"defaulting it to 30");
		nufa_buf->refresh_interval = NUFA_REFRESH_INTERVAL_DEFAULT;
	}
  
	/* Get the array built */
	while (trav_xl) {
		index++;
		trav_xl = trav_xl->next;
	}
	nufa_buf->child_count = index;
	nufa_buf->sched_index = 0;
	nufa_buf->array = CALLOC (index, sizeof (struct nufa_sched_struct));
	ERR_ABORT (nufa_buf->array);
	nufa_buf->local_array = CALLOC (index, sizeof (int32_t));
	ERR_ABORT (nufa_buf->array);
	trav_xl = xl->children;
	
	local_name = dict_get (xl->options, "scheduler.local-volume-name");
	if (!local_name) {
		/* Error */
		gf_log ("nufa", GF_LOG_ERROR, 
			"No 'local-volume-name' option given in volume file");
		FREE (nufa_buf->array);
		FREE (nufa_buf->local_array);
		FREE (nufa_buf);
		return -1;
	}

	/* Get the array properly */
	index = 0;
	trav_xl = xl->children;
	while (trav_xl) {
		nufa_buf->array[index].xl = trav_xl->xlator;
		nufa_buf->array[index].eligible = 1;
		nufa_buf->array[index].free_disk = nufa_buf->min_free_disk;
		nufa_buf->array[index].refresh_interval = 
			nufa_buf->refresh_interval;

		trav_xl = trav_xl->next;
		index++;
	}
  
	{ 
		int32_t array_index = 0;
		char *child = NULL;
		char *tmp = NULL;
		char *childs_data = strdup (local_name->data);
		
		child = strtok_r (childs_data, ",", &tmp);
		while (child) {
			/* Check if the local_volume specified is proper 
			   subvolume of unify */
			trav_xl = xl->children;
			index=0;
			while (trav_xl) {
				if (strcmp (child, trav_xl->xlator->name) == 0)
					break;
				trav_xl = trav_xl->next;
				index++;
			}

			if (!trav_xl) {
				/* entry for 'local-volume-name' is wrong, 
				   not present in subvolumes */
				gf_log ("nufa", GF_LOG_ERROR, 
					"option 'scheduler.local-volume-name' "
					"%s is wrong", child);
				FREE (nufa_buf->array);
				FREE (nufa_buf->local_array);
				FREE (nufa_buf);
				return -1;
			} else {
				nufa_buf->local_array[array_index++] = index;
				nufa_buf->local_xl_count++;
			}
			child = strtok_r (NULL, ",", &tmp);
		}
		free (childs_data);
	}

	LOCK_INIT (&nufa_buf->nufa_lock);
	*((long *)xl->private) = (long)nufa_buf; // put it at the proper place
	return 0;
}

static void
nufa_fini (xlator_t *xl)
{
	struct nufa_struct *nufa_buf = 
		(struct nufa_struct *)*((long *)xl->private);

	LOCK_DESTROY (&nufa_buf->nufa_lock);
	FREE (nufa_buf->local_array);
	FREE (nufa_buf->array);
	FREE (nufa_buf);
}

static int32_t 
update_stat_array_cbk (call_frame_t *frame,
		       void *cookie,
		       xlator_t *xl,
		       int32_t op_ret,
		       int32_t op_errno,
		       struct xlator_stats *trav_stats)
{
	struct nufa_struct *nufa_struct = NULL; 
	int32_t idx = 0;
	int32_t percent = 0;

	nufa_struct = (struct nufa_struct *)*((long *)xl->private);
	LOCK (&nufa_struct->nufa_lock);
	for (idx = 0; idx < nufa_struct->child_count; idx++) {
		if (nufa_struct->array[idx].xl->name == (char *)cookie)
			break;
	}
	UNLOCK (&nufa_struct->nufa_lock);

	if (op_ret == 0) {
		percent = ((trav_stats->free_disk * 100) / 
			   trav_stats->total_disk_size);
		if (nufa_struct->array[idx].free_disk > percent) {
			if (nufa_struct->array[idx].eligible)
				gf_log ("nufa", GF_LOG_CRITICAL,
					"node \"%s\" is _almost_ (%d %%) full",
					nufa_struct->array[idx].xl->name, 
					100 - percent);
			nufa_struct->array[idx].eligible = 0;
		} else {
			nufa_struct->array[idx].eligible = 1;
		}
	} else {
		nufa_struct->array[idx].eligible = 0;
	}

	STACK_DESTROY (frame->root);
	return 0;
}

static void 
update_stat_array (xlator_t *xl)
{
	/* This function schedules the file in one of the child nodes */
	int32_t idx;
	struct nufa_struct *nufa_buf = 
		(struct nufa_struct *)*((long *)xl->private);
	call_frame_t *frame = NULL;
	call_pool_t *pool = xl->ctx->pool;

	for (idx = 0; idx < nufa_buf->child_count; idx++) {
		frame = create_frame (xl, pool);

		STACK_WIND_COOKIE (frame,
				   update_stat_array_cbk, 
				   nufa_buf->array[idx].xl->name,
				   nufa_buf->array[idx].xl, 
				   (nufa_buf->array[idx].xl)->mops->stats,
				   0); //flag
	}

	return;
}

static void 
nufa_update (xlator_t *xl)
{
	struct nufa_struct *nufa_buf = 
		(struct nufa_struct *)*((long *)xl->private);
	struct timeval tv;
	gettimeofday (&tv, NULL);
	if (tv.tv_sec > (nufa_buf->refresh_interval 
			 + nufa_buf->last_stat_fetch.tv_sec)) {
		/* Update the stats from all the server */
		update_stat_array (xl);
		nufa_buf->last_stat_fetch.tv_sec = tv.tv_sec;
	}
}

static xlator_t *
nufa_schedule (xlator_t *xl, const void *path)
{
	struct nufa_struct *nufa_buf = 
		(struct nufa_struct *)*((long *)xl->private);
	int32_t nufa_orig = nufa_buf->local_xl_index;  
	int32_t rr;
  
	nufa_update (xl);
  
	while (1) {
		LOCK (&nufa_buf->nufa_lock);
		rr = nufa_buf->local_xl_index++;
		nufa_buf->local_xl_index %= nufa_buf->local_xl_count;
		UNLOCK (&nufa_buf->nufa_lock);
    
		/* if 'eligible' or there are _no_ eligible nodes */
		if (nufa_buf->array[nufa_buf->local_array[rr]].eligible) {
			/* Return the local node */
			return nufa_buf->array[nufa_buf->local_array[rr]].xl;
		}
		if ((rr + 1) % nufa_buf->local_xl_count == nufa_orig) {
			gf_log ("nufa", GF_LOG_CRITICAL, 
				"No free space available on any local "
				"volumes, using RR scheduler");
			LOCK (&nufa_buf->nufa_lock);
			nufa_buf->local_xl_index++;
			nufa_buf->local_xl_index %= nufa_buf->local_xl_count;
			UNLOCK (&nufa_buf->nufa_lock);
			break;
		}
	}

	nufa_orig = nufa_buf->sched_index;  
	while (1) {
		LOCK (&nufa_buf->nufa_lock);
		rr = nufa_buf->sched_index++;
		nufa_buf->sched_index = (nufa_buf->sched_index % 
					 nufa_buf->child_count);
		UNLOCK (&nufa_buf->nufa_lock);
    
		/* if 'eligible' or there are _no_ eligible nodes */
		if (nufa_buf->array[rr].eligible) {
			break;
		}
		if ((rr + 1) % nufa_buf->child_count == nufa_orig) {
			gf_log ("nufa", GF_LOG_CRITICAL, 
				"No free space available on any server, "
				"using RR scheduler.");
			LOCK (&nufa_buf->nufa_lock);
			nufa_buf->sched_index++;
			nufa_buf->sched_index = (nufa_buf->sched_index % 
						 nufa_buf->child_count);
			UNLOCK (&nufa_buf->nufa_lock);
			break;
		}
	}
	return nufa_buf->array[rr].xl;
}


/**
 * notify
 */
void
nufa_notify (xlator_t *xl, int32_t event, void *data)
{
	struct nufa_struct *nufa_buf = 
		(struct nufa_struct *)*((long *)xl->private);
	int32_t idx = 0;
  
	if (!nufa_buf)
		return;

	for (idx = 0; idx < nufa_buf->child_count; idx++) {
		if (strcmp (nufa_buf->array[idx].xl->name, 
			    ((xlator_t *)data)->name) == 0)
			break;
	}

	switch (event)
	{
	case GF_EVENT_CHILD_UP:
	{
		//nufa_buf->array[idx].eligible = 1;
	}
	break;
	case GF_EVENT_CHILD_DOWN:
	{
		nufa_buf->array[idx].eligible = 0;
	}
	break;
	default:
	{
		;
	}
	break;
	}

}

struct sched_ops sched = {
	.init     = nufa_init,
	.fini     = nufa_fini,
	.update   = nufa_update,
	.schedule = nufa_schedule,
	.notify   = nufa_notify
};

struct volume_options options[] = {
	{ .key   = { "scheduler.refresh-interval", 
		     "nufa.refresh-interval" },  
	  .type  = GF_OPTION_TYPE_TIME
	},
	{ .key   = { "scheduler.limits.min-free-disk", 
		     "nufa.limits.min-free-disk" },  
	  .type  = GF_OPTION_TYPE_PERCENT
	},
	{ .key   = { "scheduler.local-volume-name",
		     "nufa.local-volume-name" },
	  .type  = GF_OPTION_TYPE_XLATOR
	},	
	{ .key = {NULL} }
};