1 00:00:00,690 --> 00:00:02,700 In this lesson, you're going to be introduced 2 00:00:02,700 --> 00:00:05,070 to some common CPU and memory issues 3 00:00:05,070 --> 00:00:06,450 that occur within Linux, 4 00:00:06,450 --> 00:00:08,189 as well as important command tools 5 00:00:08,189 --> 00:00:09,693 that you can use to troubleshoot issues 6 00:00:09,693 --> 00:00:12,030 that you may find along the way. 7 00:00:12,030 --> 00:00:14,400 After all, sometimes unexpected behavior 8 00:00:14,400 --> 00:00:16,079 on the server side can occur 9 00:00:16,079 --> 00:00:17,010 and this can be caused 10 00:00:17,010 --> 00:00:19,680 by resource limitations on your system. 11 00:00:19,680 --> 00:00:20,880 Linux, by design, 12 00:00:20,880 --> 00:00:23,310 aims to use all of the available physical memory 13 00:00:23,310 --> 00:00:25,080 as efficiently as possible. 14 00:00:25,080 --> 00:00:28,500 But CPU and memory are two fundamental computing concepts 15 00:00:28,500 --> 00:00:30,750 and you need to be able to manage these crucial elements 16 00:00:30,750 --> 00:00:33,570 of a Linux system to ensure optimal performance 17 00:00:33,570 --> 00:00:36,420 while supporting the best user experience. 18 00:00:36,420 --> 00:00:38,640 There are many possible issues that could affect 19 00:00:38,640 --> 00:00:40,560 or be caused by the CPU. 20 00:00:40,560 --> 00:00:43,530 These issues may include an underperforming CPU 21 00:00:43,530 --> 00:00:46,110 for the desired level of speed and responsiveness. 22 00:00:46,110 --> 00:00:49,350 An overloaded CPU caused by too many requests resulting 23 00:00:49,350 --> 00:00:52,770 in a delay of response, or you can have non-functional cores 24 00:00:52,770 --> 00:00:55,290 or other things that are causing problems. 25 00:00:55,290 --> 00:00:56,730 You can use the information 26 00:00:56,730 --> 00:01:01,020 from /proc/cpuinfo to identify characteristics 27 00:01:01,020 --> 00:01:04,200 about your CPU that might indicate issues related 28 00:01:04,200 --> 00:01:06,450 to performance or a lack of support 29 00:01:06,450 --> 00:01:08,520 for specific features you might need. 30 00:01:08,520 --> 00:01:12,210 Some useful fields in the /proc/cpuinfo file 31 00:01:12,210 --> 00:01:13,530 include the processor, 32 00:01:13,530 --> 00:01:15,840 which refers to the number of logical cores, 33 00:01:15,840 --> 00:01:19,290 the vendor ID, which refers to the CPU manufacturer, 34 00:01:19,290 --> 00:01:21,750 the model name, the CPU megahertz, 35 00:01:21,750 --> 00:01:24,270 which refers to the logical core's clock speed 36 00:01:24,270 --> 00:01:27,000 and is measured out to the thousands decimal place. 37 00:01:27,000 --> 00:01:28,920 The cache size and the flags 38 00:01:28,920 --> 00:01:30,450 which refer to the characteristics 39 00:01:30,450 --> 00:01:33,840 about the CPU as well as the supported features. 40 00:01:33,840 --> 00:01:34,860 As you've seen, 41 00:01:34,860 --> 00:01:36,990 the system CTL command enables you 42 00:01:36,990 --> 00:01:39,360 to view kernel parameters at runtime. 43 00:01:39,360 --> 00:01:41,370 The uptime command will display the time 44 00:01:41,370 --> 00:01:43,800 from when the system started running as well. 45 00:01:43,800 --> 00:01:46,410 The output of the uptime command gives information 46 00:01:46,410 --> 00:01:47,670 about the current time, 47 00:01:47,670 --> 00:01:49,320 how long the system has been running, 48 00:01:49,320 --> 00:01:51,960 and how many users are currently logged in. 49 00:01:51,960 --> 00:01:54,420 Most relevant to the CPU troubleshooting, though, 50 00:01:54,420 --> 00:01:56,490 is the load average field. 51 00:01:56,490 --> 00:01:58,740 A CPU's load is going to be expressed 52 00:01:58,740 --> 00:02:00,270 as the number of processes 53 00:02:00,270 --> 00:02:03,840 that are either using or waiting to use that CPU. 54 00:02:03,840 --> 00:02:05,850 It can also include the number of processes 55 00:02:05,850 --> 00:02:08,610 in queue for storage input/output. 56 00:02:08,610 --> 00:02:11,640 Using uptime, you're going to be able to find the average load 57 00:02:11,640 --> 00:02:13,830 over three different periods of time. 58 00:02:13,830 --> 00:02:15,810 You can use these values to determine 59 00:02:15,810 --> 00:02:18,120 when your CPU has become overloaded, 60 00:02:18,120 --> 00:02:20,370 which may lead to performance issues. 61 00:02:20,370 --> 00:02:23,100 For example, if you have four logical cores 62 00:02:23,100 --> 00:02:26,580 and the load average of the last 15 minutes is 4.0, 63 00:02:26,580 --> 00:02:30,180 then that means your CPU was being fully utilized. 64 00:02:30,180 --> 00:02:32,460 If you're load average were 8.0, 65 00:02:32,460 --> 00:02:36,180 that means your CPU would be 100% overloaded. 66 00:02:36,180 --> 00:02:37,860 Now, one of the commands we can use 67 00:02:37,860 --> 00:02:42,284 to look at our system reports is known as sar, S-A-R. 68 00:02:42,284 --> 00:02:45,390 sar is going to display system usage reports based 69 00:02:45,390 --> 00:02:48,090 on data collected from system activity. 70 00:02:48,090 --> 00:02:50,610 These reports contain various different sections, 71 00:02:50,610 --> 00:02:53,310 and each of those sections consists of the type of data 72 00:02:53,310 --> 00:02:55,740 and the time at which that data was collected. 73 00:02:55,740 --> 00:02:57,780 The default mode of the sar command 74 00:02:57,780 --> 00:02:59,640 is going to display the CPU usage 75 00:02:59,640 --> 00:03:02,040 in various time increments for each category 76 00:03:02,040 --> 00:03:04,440 of resource that's accessed by the CPU. 77 00:03:04,440 --> 00:03:06,840 This includes your users, the system, 78 00:03:06,840 --> 00:03:09,210 input/output scheduling, and other things. 79 00:03:09,210 --> 00:03:11,163 It also displays the percentage of the CPU 80 00:03:11,163 --> 00:03:13,650 that was idle at any given time. 81 00:03:13,650 --> 00:03:15,750 At the bottom of the report is an average 82 00:03:15,750 --> 00:03:19,290 of each of these data points across the listed time periods. 83 00:03:19,290 --> 00:03:22,290 By default, sar reports are going to have data collected 84 00:03:22,290 --> 00:03:23,580 every 10 minutes. 85 00:03:23,580 --> 00:03:25,770 Though, you can use different options to filter 86 00:03:25,770 --> 00:03:27,690 and shape these reports as well. 87 00:03:27,690 --> 00:03:28,830 To use this command, 88 00:03:28,830 --> 00:03:31,980 simply type sar and the options you want. 89 00:03:31,980 --> 00:03:33,120 The next command we're going to talk 90 00:03:33,120 --> 00:03:37,020 about is sysctl or system control. 91 00:03:37,020 --> 00:03:38,910 The system control command can be used 92 00:03:38,910 --> 00:03:40,950 to troubleshoot CPU issues as well 93 00:03:40,950 --> 00:03:44,550 by retrieving CPU-based kernel parameters at runtime. 94 00:03:44,550 --> 00:03:47,790 One useful set of parameters concerns scheduling domains. 95 00:03:47,790 --> 00:03:49,530 Now scheduling domains are a method 96 00:03:49,530 --> 00:03:52,830 by which the kernel groups can take logical cores together 97 00:03:52,830 --> 00:03:54,420 and then share scheduling policies 98 00:03:54,420 --> 00:03:56,340 and other properties between them. 99 00:03:56,340 --> 00:03:58,560 These parameters will usually take the format 100 00:03:58,560 --> 00:04:00,813 of kernel.sched_domain.cpu# 101 00:04:03,488 --> 00:04:04,738 .domain#.param. 102 00:04:06,270 --> 00:04:07,800 Now another way to get information 103 00:04:07,800 --> 00:04:09,840 about your CPU and its architecture is 104 00:04:09,840 --> 00:04:12,090 to use the lscpu command. 105 00:04:12,090 --> 00:04:14,400 lscpu is a command line utility 106 00:04:14,400 --> 00:04:16,860 that's going to read the CPU architecture information 107 00:04:16,860 --> 00:04:20,910 from the sysfs and /proc/cpuinfo files 108 00:04:20,910 --> 00:04:22,860 and then print that to your terminal. 109 00:04:22,860 --> 00:04:24,420 The information here is going to include things 110 00:04:24,420 --> 00:04:27,270 like the number of CPUs, the threads, the cores, 111 00:04:27,270 --> 00:04:31,890 the sockets, and non-uniform memory access, or NUMA nodes. 112 00:04:31,890 --> 00:04:34,770 It's also going to be able to display CPU cache information 113 00:04:34,770 --> 00:04:38,610 and cache sharing as well as family, model, bogomips, 114 00:04:38,610 --> 00:04:41,790 byte order, and stepping of the CPU. 115 00:04:41,790 --> 00:04:43,860 Now, in addition to all the issues you can have 116 00:04:43,860 --> 00:04:47,640 with your CPU, you can also have issues caused by memory. 117 00:04:47,640 --> 00:04:49,980 Now, memory issues may include not having enough 118 00:04:49,980 --> 00:04:53,070 total memory to service all your processes at once 119 00:04:53,070 --> 00:04:55,020 or not having enough free memory 120 00:04:55,020 --> 00:04:57,630 to service new processes that need to be launched 121 00:04:57,630 --> 00:05:00,390 or processes that are unable to access memory, 122 00:05:00,390 --> 00:05:02,490 despite there being memory available 123 00:05:02,490 --> 00:05:05,280 or processes accessing too much memory 124 00:05:05,280 --> 00:05:07,200 and basically draining your system 125 00:05:07,200 --> 00:05:08,580 or you can have things 126 00:05:08,580 --> 00:05:10,950 that are leaving other processes without memory 127 00:05:10,950 --> 00:05:13,350 because they're sucking up all that extra memory. 128 00:05:13,350 --> 00:05:14,790 Or we can have a system 129 00:05:14,790 --> 00:05:17,070 that cannot quickly access files from a cache 130 00:05:17,070 --> 00:05:19,740 or a buffer because there's not enough physical memory 131 00:05:19,740 --> 00:05:22,410 and things are being put into a cache or buffer instead. 132 00:05:22,410 --> 00:05:24,780 And lots of other memory issues like this. 133 00:05:24,780 --> 00:05:26,790 Now to troubleshoot some of these issues, 134 00:05:26,790 --> 00:05:31,500 you can use information from the /proc/meminfo file. 135 00:05:31,500 --> 00:05:32,400 Now this will ensure 136 00:05:32,400 --> 00:05:34,350 that the system's RAM modules are performing 137 00:05:34,350 --> 00:05:36,240 to the specification you expect, 138 00:05:36,240 --> 00:05:37,920 and that the OS is consuming memory 139 00:05:37,920 --> 00:05:39,810 at the right rate that you expect 140 00:05:39,810 --> 00:05:42,660 and that your system has enough overall available memory 141 00:05:42,660 --> 00:05:44,640 to perform the tasks you want. 142 00:05:44,640 --> 00:05:45,930 Now, some useful fields 143 00:05:45,930 --> 00:05:50,220 in those /proc/meminfo file include things like memtotal, 144 00:05:50,220 --> 00:05:51,540 which refers to the total amount 145 00:05:51,540 --> 00:05:53,640 of physical memory in the system 146 00:05:53,640 --> 00:05:55,830 or memfree, which is the total amount 147 00:05:55,830 --> 00:05:58,110 of free memory that's available in that system 148 00:05:58,110 --> 00:05:59,430 at this time. 149 00:05:59,430 --> 00:06:01,020 Cached, this is going to refer 150 00:06:01,020 --> 00:06:02,760 to the total amount of physical memory 151 00:06:02,760 --> 00:06:04,920 that's being used as cache memory. 152 00:06:04,920 --> 00:06:07,650 Swaptotal, this refers to the total amount 153 00:06:07,650 --> 00:06:09,660 of swap space on your system. 154 00:06:09,660 --> 00:06:12,120 And swapfree, which refers to the total amount 155 00:06:12,120 --> 00:06:15,090 of swap space that is currently unused on your system. 156 00:06:15,090 --> 00:06:16,770 Dirty refers to the total amount 157 00:06:16,770 --> 00:06:19,200 of memory that is waiting to be written to storage 158 00:06:19,200 --> 00:06:21,210 and writeback refers to the total amount 159 00:06:21,210 --> 00:06:23,760 of memory currently being written to storage. 160 00:06:23,760 --> 00:06:26,100 Now the free command is another command we can use 161 00:06:26,100 --> 00:06:27,300 to look at our memory 162 00:06:27,300 --> 00:06:31,050 and free is going to parse the /proc/meminfo file 163 00:06:31,050 --> 00:06:34,500 for easier analysis for those memory usage statistics. 164 00:06:34,500 --> 00:06:37,050 Its default behavior is to display the information 165 00:06:37,050 --> 00:06:39,300 about system memory and the swap space 166 00:06:39,300 --> 00:06:41,400 as well and include the total memory, 167 00:06:41,400 --> 00:06:43,530 the total used, the total free, 168 00:06:43,530 --> 00:06:46,200 the total shared, the total buffered and cached, 169 00:06:46,200 --> 00:06:48,570 and the total available for starting new apps. 170 00:06:48,570 --> 00:06:50,370 There are several different command options 171 00:06:50,370 --> 00:06:52,350 that you can use with the free command. 172 00:06:52,350 --> 00:06:55,715 Things like -b, -k, -m, 173 00:06:55,715 --> 00:06:57,600 g, or -tera. 174 00:06:57,600 --> 00:06:59,730 And all those options will display the memory 175 00:06:59,730 --> 00:07:03,420 in either bytes, kilobytes, megabytes, gigabytes, 176 00:07:03,420 --> 00:07:05,430 or terabytes, respectively. 177 00:07:05,430 --> 00:07:08,370 If you use -s and the number of seconds you want, 178 00:07:08,370 --> 00:07:09,570 it will tell you how quickly 179 00:07:09,570 --> 00:07:11,790 you want to update the memory statistics based 180 00:07:11,790 --> 00:07:13,500 on the number of seconds you tell it to. 181 00:07:13,500 --> 00:07:17,524 So you might say -s10 to do it every 10 seconds. 182 00:07:17,524 --> 00:07:19,680 o is going to disable the display 183 00:07:19,680 --> 00:07:22,350 of the buffered or cached information. 184 00:07:22,350 --> 00:07:24,540 t is going to display the total line 185 00:07:24,540 --> 00:07:26,220 that's going to combine physical memory 186 00:07:26,220 --> 00:07:29,610 with swap space as one combined memory area. 187 00:07:29,610 --> 00:07:33,420 And -h will give you output that is more human readable. 188 00:07:33,420 --> 00:07:35,100 Now memory can also be cached 189 00:07:35,100 --> 00:07:37,230 and this means it's going to be stored temporarily 190 00:07:37,230 --> 00:07:39,570 so that the data can then be accessed more quickly 191 00:07:39,570 --> 00:07:40,770 in the future. 192 00:07:40,770 --> 00:07:41,910 Now, the buffers field 193 00:07:41,910 --> 00:07:46,050 inside the /proc/meminfo file indicates memory 194 00:07:46,050 --> 00:07:48,390 that has been assigned to a specific block device, 195 00:07:48,390 --> 00:07:49,800 like a hard drive. 196 00:07:49,800 --> 00:07:53,070 This memory can then be used to cache file system metadata 197 00:07:53,070 --> 00:07:54,960 like directory contents, permissions, 198 00:07:54,960 --> 00:07:56,610 and other things like that. 199 00:07:56,610 --> 00:07:58,800 The cached memory is going to be similar to this 200 00:07:58,800 --> 00:08:01,140 but instead of storing file metadata, 201 00:08:01,140 --> 00:08:04,020 it's going to store the actual contents of the file. 202 00:08:04,020 --> 00:08:06,420 So the buffers store metadata 203 00:08:06,420 --> 00:08:09,390 but cache is going to store the actual files. 204 00:08:09,390 --> 00:08:11,190 Now, when you use the free command, 205 00:08:11,190 --> 00:08:13,290 we can combine these two values together 206 00:08:13,290 --> 00:08:15,480 inside of our output as well. 207 00:08:15,480 --> 00:08:17,040 Another way to look at the information 208 00:08:17,040 --> 00:08:20,010 about your memory is to use the lsmem command. 209 00:08:20,010 --> 00:08:22,110 The lsmem command is going to list the ranges 210 00:08:22,110 --> 00:08:24,780 of available memory with our online status. 211 00:08:24,780 --> 00:08:26,610 This will show you all the different memory blocks 212 00:08:26,610 --> 00:08:30,750 that correspond to the memory blocks inside of the sysfs. 213 00:08:30,750 --> 00:08:33,059 This command will also show the memory block size, 214 00:08:33,059 --> 00:08:34,049 the device size, 215 00:08:34,049 --> 00:08:37,023 and the amount of memory in online or offline statuses. 216 00:08:37,980 --> 00:08:40,860 Another command we need to talk about is vmstat. 217 00:08:40,860 --> 00:08:44,039 Now vmstat is a command that displays various statistics 218 00:08:44,039 --> 00:08:45,630 about virtual memory. 219 00:08:45,630 --> 00:08:48,390 This includes things like our processes, our CPU, 220 00:08:48,390 --> 00:08:50,940 and our IO statistics, as well as our virtual memory 221 00:08:50,940 --> 00:08:52,740 when we use vmstat. 222 00:08:52,740 --> 00:08:55,200 Now by default, the report that you're going to get 223 00:08:55,200 --> 00:08:57,570 from vmstat is going to provide you with averages 224 00:08:57,570 --> 00:09:00,720 for each statistic since the last system boot up. 225 00:09:00,720 --> 00:09:03,390 But you can also specify a delay value 226 00:09:03,390 --> 00:09:05,190 to sample from a period of time. 227 00:09:05,190 --> 00:09:08,340 For instance, every 60 seconds, you want to get a new report, 228 00:09:08,340 --> 00:09:10,020 you can tell it to do that. 229 00:09:10,020 --> 00:09:11,760 Now, some memory-based statistics here 230 00:09:11,760 --> 00:09:14,250 are going to include total virtual memory available. 231 00:09:14,250 --> 00:09:15,360 The total virtual memory 232 00:09:15,360 --> 00:09:16,800 that's free for us to use, 233 00:09:16,800 --> 00:09:19,230 the total memory used in the buffers and cache, 234 00:09:19,230 --> 00:09:21,870 and the total memory that's used in swap spaces. 235 00:09:21,870 --> 00:09:23,670 Some CPU-based statistics 236 00:09:23,670 --> 00:09:26,370 are going to include time spent running user space, 237 00:09:26,370 --> 00:09:28,560 the time spent running in kernel space, 238 00:09:28,560 --> 00:09:31,050 the time spent idle, and the time spent waiting 239 00:09:31,050 --> 00:09:32,580 for input and output. 240 00:09:32,580 --> 00:09:35,700 To run the vmstat command, type in vmstat, 241 00:09:35,700 --> 00:09:39,360 the options, the delay with the counter, and then hit enter. 242 00:09:39,360 --> 00:09:41,340 It's recommended that you supply vmstat 243 00:09:41,340 --> 00:09:44,130 with a delay for more accurate report as well. 244 00:09:44,130 --> 00:09:48,240 For example, vmstat five five will run the command 245 00:09:48,240 --> 00:09:51,210 on a five-second delay for five different intervals, 246 00:09:51,210 --> 00:09:53,820 giving you a good average over that time. 247 00:09:53,820 --> 00:09:56,460 Another thing we want to talk about is OOM, 248 00:09:56,460 --> 00:09:59,340 the out of memory killer or OOM is a feature 249 00:09:59,340 --> 00:10:00,960 of the Linux kernel that determines 250 00:10:00,960 --> 00:10:02,940 what processes need to be killed 251 00:10:02,940 --> 00:10:05,790 when the system becomes extremely low on memory. 252 00:10:05,790 --> 00:10:08,850 The out of memory killer will continue to kill processes 253 00:10:08,850 --> 00:10:10,830 until enough free memory is made 254 00:10:10,830 --> 00:10:11,663 so that the kernel 255 00:10:11,663 --> 00:10:14,220 and the system can begin running smoothly again. 256 00:10:14,220 --> 00:10:16,710 Rather than killing processes at random though, 257 00:10:16,710 --> 00:10:19,560 the out of memory killer will leverage an algorithm 258 00:10:19,560 --> 00:10:22,650 that assigns each process an OOM score. 259 00:10:22,650 --> 00:10:23,940 The higher the score, 260 00:10:23,940 --> 00:10:26,670 the higher the chance that process has of being killed 261 00:10:26,670 --> 00:10:28,680 during an out of memory event. 262 00:10:28,680 --> 00:10:30,480 The assignment algorithm is going to consider 263 00:10:30,480 --> 00:10:32,910 what processes will free up the greatest amount of memory 264 00:10:32,910 --> 00:10:36,000 when killed as well as what processes are least important 265 00:10:36,000 --> 00:10:37,560 to system stability. 266 00:10:37,560 --> 00:10:39,480 Then it'll assign a score based 267 00:10:39,480 --> 00:10:41,970 on what it determines to be the most optimal targets 268 00:10:41,970 --> 00:10:43,260 for termination. 269 00:10:43,260 --> 00:10:46,170 Although this mechanism is mostly automated by the kernel, 270 00:10:46,170 --> 00:10:49,230 you do have the ability to control some of its behavior. 271 00:10:49,230 --> 00:10:51,300 You can mount the OOM control group 272 00:10:51,300 --> 00:10:53,070 as a desired mount point. 273 00:10:53,070 --> 00:10:54,270 And if, for example 274 00:10:54,270 --> 00:10:57,420 you want to designate a group of processes to be killed first, 275 00:10:57,420 --> 00:10:59,760 you can create a directory at this mount point 276 00:10:59,760 --> 00:11:02,070 and create a task file in this directory 277 00:11:02,070 --> 00:11:05,250 then list out the PIDs of the relevant processes. 278 00:11:05,250 --> 00:11:08,760 Then you can create an OOM.priority file 279 00:11:08,760 --> 00:11:11,880 with a higher integer value like 256 280 00:11:11,880 --> 00:11:14,010 to give the processes a higher priority 281 00:11:14,010 --> 00:11:16,590 for the OOM killer to take care of. 282 00:11:16,590 --> 00:11:19,560 Or to prevent a process from being killed entirely, 283 00:11:19,560 --> 00:11:22,740 you can assign it zero as a priority value. 284 00:11:22,740 --> 00:11:25,320 Now the configuration of swap space is another thing we need 285 00:11:25,320 --> 00:11:26,153 to think about 286 00:11:26,153 --> 00:11:27,240 because your swap space 287 00:11:27,240 --> 00:11:29,730 can help alleviate memory-related issues, 288 00:11:29,730 --> 00:11:31,260 especially when your system 289 00:11:31,260 --> 00:11:33,360 and applications request more memory 290 00:11:33,360 --> 00:11:35,460 than your system physically has. 291 00:11:35,460 --> 00:11:36,780 Now systems with a low amount 292 00:11:36,780 --> 00:11:38,520 of memory are particularly vulnerable 293 00:11:38,520 --> 00:11:40,260 to these issues of swapping. 294 00:11:40,260 --> 00:11:43,470 Swap space is not a replacement for adding more memory 295 00:11:43,470 --> 00:11:45,600 but it can help minimize system 296 00:11:45,600 --> 00:11:49,590 and application sluggishness, unresponsiveness, and crashes. 297 00:11:49,590 --> 00:11:52,410 Swap space can be of one of three types. 298 00:11:52,410 --> 00:11:54,420 A device swap space is configured 299 00:11:54,420 --> 00:11:56,190 when you partition the storage device 300 00:11:56,190 --> 00:11:59,220 to allow you to have some swap space on that disc. 301 00:11:59,220 --> 00:12:01,080 This will be used by the operating system 302 00:12:01,080 --> 00:12:03,030 to run larger applications. 303 00:12:03,030 --> 00:12:05,610 File system swap space is configured primarily 304 00:12:05,610 --> 00:12:07,230 when you install Linux. 305 00:12:07,230 --> 00:12:09,120 It's going to be used by the operating system 306 00:12:09,120 --> 00:12:10,680 as an emergency resource 307 00:12:10,680 --> 00:12:13,230 when the available swap space runs out. 308 00:12:13,230 --> 00:12:14,220 On the other hand, 309 00:12:14,220 --> 00:12:16,980 pseudo swap space enables large applications 310 00:12:16,980 --> 00:12:19,590 to run on computers with limited RAM. 311 00:12:19,590 --> 00:12:22,290 Swap files are created for storing data 312 00:12:22,290 --> 00:12:23,160 that's going to be transferred 313 00:12:23,160 --> 00:12:25,860 from a system's memory to a storage device. 314 00:12:25,860 --> 00:12:28,290 These things are dynamic and they change in size 315 00:12:28,290 --> 00:12:30,690 when the data's moved in and out of memory. 316 00:12:30,690 --> 00:12:33,120 Swap files are going to be used as a medium 317 00:12:33,120 --> 00:12:36,720 to transfer data from RAM into that storage device. 318 00:12:36,720 --> 00:12:39,450 A swap partition is an area of virtual memory 319 00:12:39,450 --> 00:12:42,390 on a storage device that complement the physical RAM 320 00:12:42,390 --> 00:12:43,740 in the computer. 321 00:12:43,740 --> 00:12:45,930 Swap partitions are used by Linux 322 00:12:45,930 --> 00:12:48,930 because they perform better than a swap file system. 323 00:12:48,930 --> 00:12:50,610 Now, if you want to create a swap space 324 00:12:50,610 --> 00:12:52,050 on a storage partition, 325 00:12:52,050 --> 00:12:54,360 you're going to use the make swap command 326 00:12:54,360 --> 00:12:57,090 which is M-K-S-W-A-P. 327 00:12:57,090 --> 00:12:59,970 The make swap command is going to be used to create swap space 328 00:12:59,970 --> 00:13:02,280 on a storage partition and is typically used 329 00:13:02,280 --> 00:13:05,340 when you want to move swap space to a different partition 330 00:13:05,340 --> 00:13:06,780 than the one it was originally created 331 00:13:06,780 --> 00:13:09,000 on during system installation. 332 00:13:09,000 --> 00:13:11,460 For example, let's say you wanted to save space 333 00:13:11,460 --> 00:13:13,500 on a low capacity boot drive. 334 00:13:13,500 --> 00:13:15,240 You could provide an option to be able 335 00:13:15,240 --> 00:13:18,180 to move that over to a new swap space. 336 00:13:18,180 --> 00:13:19,620 If you use -c, 337 00:13:19,620 --> 00:13:21,600 this option verifies that the device is free 338 00:13:21,600 --> 00:13:24,180 from bad sectors before mounting the swap space. 339 00:13:24,180 --> 00:13:26,223 Essentially it's a check disc. 340 00:13:26,223 --> 00:13:28,410 p is going to set the page size 341 00:13:28,410 --> 00:13:30,690 that's going to be used by the make swap command. 342 00:13:30,690 --> 00:13:34,530 And -l will activate the swap space using labels applied 343 00:13:34,530 --> 00:13:36,750 to partitions or file systems. 344 00:13:36,750 --> 00:13:39,720 Another command we have is known as swap on. 345 00:13:39,720 --> 00:13:43,260 The swap on command is used to activate a swap partition 346 00:13:43,260 --> 00:13:44,700 in a specified device. 347 00:13:44,700 --> 00:13:46,620 While the swap off command is used 348 00:13:46,620 --> 00:13:49,500 to deactivate the swap space on a device. 349 00:13:49,500 --> 00:13:51,300 Some of the frequently used swap on 350 00:13:51,300 --> 00:13:54,630 and swap off command options are swapon-e, 351 00:13:54,630 --> 00:13:57,360 which is used to skip devices that do not exist 352 00:13:57,360 --> 00:14:01,230 or swapon-a, which is used to activate all of the swap space 353 00:14:01,230 --> 00:14:04,260 or swapoff-a, which is just to deactivate all 354 00:14:04,260 --> 00:14:05,640 of your swap space. 355 00:14:05,640 --> 00:14:07,230 All right, so now that you're familiar with some 356 00:14:07,230 --> 00:14:09,510 of the basic command tools that you can use, 357 00:14:09,510 --> 00:14:12,060 including what they do and how they can help you, 358 00:14:12,060 --> 00:14:14,160 you're now going to be able to troubleshoot CPU 359 00:14:14,160 --> 00:14:15,600 and memory issues. 360 00:14:15,600 --> 00:14:17,760 When you're troubleshooting CPU and memory issues, 361 00:14:17,760 --> 00:14:21,390 remember, identify key information about the CPU 362 00:14:21,390 --> 00:14:26,250 and its logical cores using the /proc/cpuinfo file. 363 00:14:26,250 --> 00:14:29,670 Use the uptime command to identify CPU load averages 364 00:14:29,670 --> 00:14:30,840 and use the sar command 365 00:14:30,840 --> 00:14:34,170 to see what components are causing heavy load on your CPU. 366 00:14:34,170 --> 00:14:35,700 When you want to look at memory, 367 00:14:35,700 --> 00:14:38,550 you can identify key information about your memory usage 368 00:14:38,550 --> 00:14:41,700 by looking at the /proc/meminfo file, 369 00:14:41,700 --> 00:14:43,560 or you can use the free command 370 00:14:43,560 --> 00:14:46,200 to more easily analyze memory usage information 371 00:14:46,200 --> 00:14:48,990 and the vmstat command to retrieve more information 372 00:14:48,990 --> 00:14:51,270 on both your CPU and memory usage, 373 00:14:51,270 --> 00:14:53,190 as well as your virtual memory. 374 00:14:53,190 --> 00:14:56,730 Also, you may want to consider tweaking your OOM killer 375 00:14:56,730 --> 00:14:59,730 to spare or sacrifice specific processes 376 00:14:59,730 --> 00:15:01,320 when you're running low on memory. 377 00:15:01,320 --> 00:15:02,160 And finally, 378 00:15:02,160 --> 00:15:04,200 remember that you can always create more swap space 379 00:15:04,200 --> 00:15:06,000 if you need to because you ran out 380 00:15:06,000 --> 00:15:08,463 of physical memory by using your swap commands.