1 00:00:02,290 --> 00:00:03,750 We'll come back with more. 2 00:00:03,770 --> 00:00:06,220 Or just killing your best practice questions. 3 00:00:06,230 --> 00:00:08,890 And here is another one. 4 00:00:09,020 --> 00:00:15,680 You are the best team lead and you can do it too incorporated a member of your team wants to troubleshoot 5 00:00:15,830 --> 00:00:19,920 a problem on an easy to instance launched by an older killing group. 6 00:00:21,160 --> 00:00:28,400 So you have one of the two instances launched by your group and one of that and one of your team members. 7 00:00:28,690 --> 00:00:32,120 He wants to troubleshoot this easy to instance for a specific problem. 8 00:00:32,140 --> 00:00:32,980 So what's the problem. 9 00:00:32,980 --> 00:00:39,670 The problem is causing auto scaling group to launch new instances based on a dynamic scaling policy 10 00:00:39,720 --> 00:00:44,190 what's known as calling top policy that's the on demand that's even based. 11 00:00:44,710 --> 00:00:49,960 You want to make sure that all killing will not continue to launch additional instances because of this 12 00:00:50,080 --> 00:00:53,990 or due to some troubleshooting planned activities. 13 00:00:54,010 --> 00:01:00,650 How can you carry out the required troubleshooting while ensuring it's getting will not trigger alarms 14 00:01:01,940 --> 00:01:07,530 and launch new instances and you have to choose two possible ways to do that. 15 00:01:07,770 --> 00:01:12,140 Eg move the instance into standby mode and carry out the troubleshooting. 16 00:01:12,140 --> 00:01:18,050 And you know this one is correct because usually under the skin in Group A healthy instance is yes it's 17 00:01:18,140 --> 00:01:19,590 in service. 18 00:01:19,850 --> 00:01:26,570 If you take it to stand by you what happens it will still be under the umbrella of the orders killing 19 00:01:26,570 --> 00:01:30,450 groups a lot of killing is still encompassing these. 20 00:01:30,500 --> 00:01:36,630 But it will be doing your health status on health checks on the inservice ones and not on the stand 21 00:01:36,640 --> 00:01:43,520 by which means you can stop it restart it installs things stock applications do whatever you like without 22 00:01:43,760 --> 00:01:46,770 the auto scanning determining that this one is unhealthy. 23 00:01:46,820 --> 00:01:48,940 We have to terminate it and launch a new one. 24 00:01:49,080 --> 00:01:50,920 You cannot insure that. 25 00:01:50,950 --> 00:01:51,700 And this is wrong. 26 00:01:51,710 --> 00:01:52,170 See. 27 00:01:52,190 --> 00:01:56,900 Otis King will continue to launch two instances once you start your troubleshooting activities. 28 00:01:56,930 --> 00:01:58,480 Well it depends on how you do it. 29 00:01:58,480 --> 00:02:00,700 So this is too generic of a statement. 30 00:02:01,280 --> 00:02:05,480 You can suspend all the screening process temporarily and this is also correct. 31 00:02:05,480 --> 00:02:11,610 This is a fact that you can suspend doctors can process them temporarily just analysis. 32 00:02:11,810 --> 00:02:14,380 You can do this for any. 33 00:02:14,450 --> 00:02:19,160 You can move the instance into standby carry out the troubleshooting and then put it back once done 34 00:02:20,950 --> 00:02:23,560 B is incorrect. 35 00:02:23,900 --> 00:02:31,250 You can insure that C is incorrect because you can stop all the scaling from considering that you're 36 00:02:31,310 --> 00:02:36,920 easy to instances that you are troubleshooting is unhealthy and launching new instances and D is correct 37 00:02:36,950 --> 00:02:39,950 because you can do that you can suspend all this killing group. 38 00:02:39,980 --> 00:02:48,970 And the correct answers are A and B moving an instance from inservice to stung by state keeps it managed 39 00:02:48,970 --> 00:02:50,490 by the authors getting process. 40 00:02:50,500 --> 00:02:56,200 But it does not count towards the active instances or just getting process will stop cheeking for the 41 00:02:56,200 --> 00:03:02,830 health of the instances that are in standby and that allows you to troubleshoot stop start do whatever 42 00:03:02,830 --> 00:03:04,020 you want. 43 00:03:04,600 --> 00:03:09,940 And this is important getting troubleshooting or making changes to your easy to instances without having 44 00:03:10,030 --> 00:03:16,780 to scan and consider them as unhealthy because of trigger events or alarms. 45 00:03:16,800 --> 00:03:21,870 You can also suspend the auto screening process for some time if you have or need to but you need to 46 00:03:21,870 --> 00:03:28,980 keep in mind that it no longer can scale out or skill and so the longer that period you are suspending 47 00:03:28,980 --> 00:03:36,270 it means you are jeopardizing or putting your workload and production environment at risk because it 48 00:03:36,270 --> 00:03:44,250 cannot scale out what's killing scale out means it can impact the availability of your services skill 49 00:03:44,270 --> 00:03:49,620 and if it cannot screen that means you have more to insist that you don't need to be physically more 50 00:03:49,620 --> 00:03:52,140 cost. 51 00:03:52,160 --> 00:03:58,730 Another question your administrator not just that one of the instances launched by auto scaling process 52 00:03:58,850 --> 00:04:01,810 is showing a status of impaired. 53 00:04:01,910 --> 00:04:07,430 So one of the instances under your auto screening group that was launched on the largest killing group 54 00:04:07,430 --> 00:04:09,320 is shown a status of impaired. 55 00:04:09,320 --> 00:04:10,700 What does that mean. 56 00:04:10,760 --> 00:04:18,500 That means it has detected a hardware or software problem on the physical host where the easy to instance 57 00:04:18,510 --> 00:04:27,360 exists and is willing to terminate or stop the instance depending on the type of the bookstore that 58 00:04:27,360 --> 00:04:28,560 you have. 59 00:04:28,560 --> 00:04:35,690 And there is a status check process will report this to auto scaling process. 60 00:04:35,780 --> 00:04:37,580 So it is impaired. 61 00:04:37,660 --> 00:04:43,730 And what will happen next when it is time for the easy to status easy to service to send to a status 62 00:04:43,730 --> 00:04:46,570 check to go to scaling. 63 00:04:46,670 --> 00:04:48,200 It will inform it. 64 00:04:48,230 --> 00:04:49,660 That is no longer an issue. 65 00:04:49,670 --> 00:04:58,820 It is impaired What action would all descaling take one's status checks report the instance status as 66 00:04:58,820 --> 00:05:03,260 impaired to what would auto Skilling take once they're easy to. 67 00:05:03,260 --> 00:05:09,560 So once that is reported to auto scaling What is the action taken by auto scaling it will launch a new 68 00:05:09,560 --> 00:05:15,890 instance immediately then work the impaired one for replacement B it will mark the instance for termination 69 00:05:15,890 --> 00:05:18,280 terminated then launch a replacement. 70 00:05:18,320 --> 00:05:24,860 See it will do not do much the existing instances under that water skiing group can take the hit of 71 00:05:24,860 --> 00:05:31,370 losing one instance the easy to statistics do not count towards the health of the Skilling launched 72 00:05:31,460 --> 00:05:32,450 instances. 73 00:05:32,510 --> 00:05:37,730 Of course this is wrong because that's the mean or default mechanism for getting to know about the health 74 00:05:38,150 --> 00:05:40,490 of the outer getting group instances. 75 00:05:40,790 --> 00:05:41,660 It will not do much. 76 00:05:41,660 --> 00:05:50,260 This is also wrong because auto scaling is all about taking care of two instances increasing decreasing 77 00:05:50,290 --> 00:05:52,980 scaling out getting in based on the need. 78 00:05:53,330 --> 00:06:00,290 It will mark the instance for them and so now that the answer is between any here it says it will launch 79 00:06:00,290 --> 00:06:06,050 in your instance immediately then mark the impaired for replacement. 80 00:06:06,150 --> 00:06:12,660 Besides it will more of instance for termination terminated then launch a replacement and B is the correct 81 00:06:12,660 --> 00:06:12,930 one. 82 00:06:12,930 --> 00:06:13,710 Why. 83 00:06:13,710 --> 00:06:18,430 Because you need to get rid of the bad apples first and been replaced with a good apple. 84 00:06:18,660 --> 00:06:19,830 That's how it works. 85 00:06:19,830 --> 00:06:25,860 But Apple is not helping you because it will not be serving your production traffic and then you get 86 00:06:25,860 --> 00:06:26,730 a new one. 87 00:06:26,730 --> 00:06:31,090 If it was rebalancing that's another story because you have all good apples. 88 00:06:31,080 --> 00:06:36,560 I can take one from you until I give you one and then I take the one I want. 89 00:06:36,570 --> 00:06:37,890 Right. 90 00:06:38,100 --> 00:06:40,320 So B is the correct answer. 91 00:06:42,480 --> 00:06:47,700 And the differential for this question by default or just getting use easy to status checks to determine 92 00:06:47,700 --> 00:06:50,670 the status of its launched etc. and since that's the default. 93 00:06:50,670 --> 00:06:58,560 What else can I have the Albee's if I choose to combine easy two plus Elby or Elby depending on how 94 00:06:58,560 --> 00:07:05,340 many I define and there that are just getting grouped any status for the two instances reported by the 95 00:07:05,340 --> 00:07:08,760 easy to status checks other than running. 96 00:07:09,120 --> 00:07:14,610 Will it mean that other scaling will mark the instance for replacement terminated and then launch a 97 00:07:14,730 --> 00:07:15,590 new one. 98 00:07:19,650 --> 00:07:26,300 If areas determines a hardware or software problem or on a host it will mark all instances with a status 99 00:07:26,310 --> 00:07:28,880 impaired and schedule them for termination. 100 00:07:28,890 --> 00:07:31,760 Or can we stop them. 101 00:07:31,790 --> 00:07:36,280 So the ones that have abs backed data stores. 102 00:07:36,300 --> 00:07:37,050 Right. 103 00:07:37,190 --> 00:07:45,440 So next one you have designed a highly available A.W. solution for a two tier web application across 104 00:07:45,440 --> 00:07:46,870 two availabilities zones. 105 00:07:47,090 --> 00:07:52,590 So you have a two tiered web application across to every meeting on of all of these on one every bit 106 00:07:52,590 --> 00:07:54,230 is on two and a half here. 107 00:07:55,290 --> 00:08:00,780 Web slush up and a have here with slush and then what do I have else in there. 108 00:08:00,790 --> 00:08:07,540 It is on a database and it the debates and that's what we call a two tier web app in one layer and it 109 00:08:07,740 --> 00:08:12,670 in another layer using that group and then he'll be to balance the front end with zero. 110 00:08:12,680 --> 00:08:20,900 So there is also an auto scaling group and just getting loop has a load balancer defined to load balance 111 00:08:20,990 --> 00:08:25,200 among the Web or the player when deployed. 112 00:08:25,270 --> 00:08:30,860 Administrators noticed that one is he does not have any easy two instances your administrator noticed 113 00:08:30,860 --> 00:08:33,970 that there are no easy to instances whatsoever in one area. 114 00:08:33,980 --> 00:08:40,500 It is only to say easy one launched by the oldest killing group checking the instances health checking 115 00:08:40,520 --> 00:08:42,180 the instance health. 116 00:08:42,180 --> 00:08:47,920 He could not find any issues with easy instances in the Skilling group or he will be so all descaling 117 00:08:48,010 --> 00:08:54,550 or he'll be both are showing the registered instances as healthy. 118 00:08:54,990 --> 00:08:59,450 So now why I do not have two instances here. 119 00:08:59,620 --> 00:09:06,400 What would be your next step towards finding out the root cause why I dont have the scaling initiated 120 00:09:06,400 --> 00:09:13,220 is it two instances in these really beautiful a chick that the plastic look balancer has a public subnet 121 00:09:13,220 --> 00:09:17,630 defined in each of the cities on this one is wrong. 122 00:09:17,650 --> 00:09:23,010 Why have you think about it for two or three seconds. 123 00:09:23,020 --> 00:09:27,660 This is saying check that the elastic like balancer has a public subnet defined in each of them it is 124 00:09:27,660 --> 00:09:31,990 on to find out whether this one is relevant or not what is the problem. 125 00:09:32,050 --> 00:09:36,030 The problem is I dont have any easy to instances in these I believe it is on. 126 00:09:36,050 --> 00:09:38,000 So why should I care if they will be. 127 00:09:38,260 --> 00:09:42,930 So now why should I now care if they will be a public servant here or not. 128 00:09:43,180 --> 00:09:44,500 I dont have any sequences. 129 00:09:44,530 --> 00:09:51,960 If the question was you have easy to incenses it is on one but they are not receiving any traffic. 130 00:09:51,960 --> 00:09:58,840 Then I would think about a good balancer but I dont have easy two instances so I should worry about 131 00:09:58,870 --> 00:10:00,450 having two instances. 132 00:10:00,460 --> 00:10:05,040 Then I monitor traffic and see if they will be sending them traffic or not. 133 00:10:05,290 --> 00:10:06,650 So thats why this one is wrong. 134 00:10:07,680 --> 00:10:12,500 We check that the auto scanning group is defined in each availabilities. 135 00:10:13,260 --> 00:10:20,820 So this one is suggesting maybe you did not define subnets in these availabilities on under the old 136 00:10:20,820 --> 00:10:25,140 Terschelling group and this is a potential correct answer. 137 00:10:25,310 --> 00:10:31,410 See the other two See create a new launch configuration with detailed monitoring on in-situ instances. 138 00:10:31,410 --> 00:10:33,150 Again this has nothing to do with monitoring. 139 00:10:33,150 --> 00:10:37,540 I dont have two instances in everybodys on one at all. 140 00:10:37,620 --> 00:10:38,990 So its not about monitoring. 141 00:10:39,090 --> 00:10:40,390 Thats why this is wrong. 142 00:10:40,400 --> 00:10:43,940 The you cannot use you will be with auto scaling. 143 00:10:43,950 --> 00:10:51,280 They are not compatible services so you can't use plastic load balancer with auto scaling. 144 00:10:51,450 --> 00:10:55,680 I can use the albuterol skinning group and that's what we have been talking about for hours and hours 145 00:10:55,680 --> 00:11:00,680 so be is your correct answer is incorrect because the problem is not with ill be covering both. 146 00:11:00,680 --> 00:11:04,660 Is it is that no instance in one of the two limit is. 147 00:11:05,110 --> 00:11:10,140 B is correct because all this killing is a service responsible for launching or tell me anything instances 148 00:11:11,110 --> 00:11:17,220 see is incorrect because this has nothing to do with details versus basic monitoring and these incorrect 149 00:11:17,220 --> 00:11:18,740 because they can be used together. 150 00:11:18,750 --> 00:11:19,340 Ill be. 151 00:11:19,350 --> 00:11:26,380 And or just get in group and the correct answer is B quick refresher or just getting groups are specific 152 00:11:26,380 --> 00:11:32,600 to adhesion and killing Skilling group can be defined in multiple availability zones but you have to 153 00:11:32,600 --> 00:11:39,740 select manually or if you have automation tools you have to define subnets in that available zone one 154 00:11:39,740 --> 00:11:47,420 or more in order to define what the Skilling group in that it is on and skin can kill out or in in those 155 00:11:47,430 --> 00:11:48,210 I believe it is on. 156 00:11:48,250 --> 00:11:48,650 All right. 157 00:11:48,650 --> 00:11:51,760 So let's take a break and I will see you in the next lecture. 158 00:11:51,770 --> 00:11:57,860 And still with all the in group scenario based practice questions I will see them.