= Monitor Performance :toc: :icons: :linkattrs: :imagesdir: ../../resources/images == Summary This section will monitor the performance of *Amazon FSx for Lustre*. == Duration NOTE: It will take approximately 5 minutes to complete this section. == Step-by-step Guide === CloudWatch dashboard . *_Open_* the link:https://console.aws.amazon.com/fsx/[Amazon FSx] console and *_select_* the link of the *File system Name* or *File system ID*. . *_Click_* the clipboard copy button next to the filesystem ID. . *_Scroll down_* and *_select_* the *Monitoring* tab. . *_Scroll down_* and *_select_* *Add to dashboard*. . From the *Add to dashboard* window, *_select_* *Create new*. . In the *Create new dashboard* text field, *paste* the *File system ID* (e.g. fs-0f19b0273cce46a41) value you copied earlier then *_select_* *Create*. . After a few seconds the new dashboard will be created. *_Select_* *Add to dashboard*. . *_Select_* *Save dashboard*. . Make sure you are using the *new interface*. *_Click_* the *Try out the new interface* link at the top right of the dashboard window if available. . *_Select_* *Add widget* then the widget type *Line* then *Metrics* from the *Add widget* window. . In the *Metrics* search text field, *_paste_* the *File system ID* (e.g. fs-0f19b0273cce46a41) value you copied earlier. . *_Scroll down_* and *_select_* the *FSx > File System Metrics 8* link. This is a link to the metrics for the file system ID. . *_Examine_* all the metrics available for the file system. . *_Click_* the *check boxes* next to the *Metric Names* *PhysicalDiskUsage* and *LogicalDiskUsage* then *_click_* *Create widget*. . Expand the width of the *LogicalDiskUsage, PhysicalDiskUsage* widget. *_Select_* the expand widget anchor (small triangle at the bottom right corning of the widget) and *_drag_* it all the way to the right side of the window. . From the *CloudWatch Dashboard* window, *_scroll up_* and *_click_* *Save dashboard*. . *_Move_* the cursor over the widget to *_examine_* and *_compare_* the difference between the physical and logical sizes of the file system over time. === CloudWatch alarm IMPORTANT: Read through all steps below before *_creating_* the alarm. . From *Total throughput (bytes/sec)* widget of the dashboard, *_click_* the *Widget actions* link image:widget-actions.jpg[align="left",width=20] then *_click_* *View in Metrics* at the bottom of the pop-up actions menu. . *_Move_* the cursor over the graph to *_examine_* the metric math (calculated) *Total throughput* metric over time. . *_Find_* the metric labeled *Total throughput* and *_click_* the *Create alarm* image:create-alarm.jpg[align="left",width=20] button in the *Actions* column. . *_Scroll_* down to the *Conditions* section and in the *Define the threshold value* field *_enter_* 1000000000. This will set an alarm condition that will trigger if the *Total Data Throughput (B/s)* is greater than 1000 MB/s. . *_Click_* *Next*. . In the *Notification* section, *_select_* *Create a new topic...* then *_enter_* an email address to which you have immediate access. . *_Click_* *Create topic*. . *_Scroll down_* and *_select_* *Next*. . Give the alarm a name. *_Enter_* a name of the alarm. e.g. High throughput. . *_Click_* *Next*. . From the *Preview and create* page, *_scroll down_* and *_click_* *Create alarm*. . You must confirm your email subscription to the SNS topic you created for this alarm. *_Access_* the inbox of the email address you entered. It should have a new email message with a subject *AWS Notification - Subscription Confirmation* from *AWS Notifications *. Read the email and *_click_* the *Confirm subscription* link. TIP: SNS topic confirmation can be done on your laptop or mobile device. . Open one (1) *Session Manager* terminal window connected to the *Linux Instance 0*. . Use ior to generate continuous load against the file system. *_Run_* the mpirun ior script below in *Session Manager* terminal window. + [source,bash] ---- _job_name=ior-performance _segment_count=32768 _threads=144 _path=/fsx/${_job_name} mkdir -p ${_path} cd /fsx mpirun --npernode ${_threads} --oversubscribe ior --posix.odirect -t 1m -b 1m -s ${_segment_count} -g -v -w -i 10 -F -k -D 0 -o ${_path}/ior.bin ---- + . While the *mpirun ior* script is running, return to the CloudWatch dashboard and *_monitor_* performance of the file system. . *_Enable_* auto-refresh of all the dashboard widgets. *_Click_* the auto-fresh button image:auto-refresh.jpg[align="left",width=20] and *_select_* *1m* to automatically refresh the dashboard every minute. Monitor the file system performance for ~5 minutes. * How much throughput is the *mpirun ior* script generating? * What is happening to the free storage capacity and disk usage metrics? * Did your *Total throughput alarm* get triggered? * *_Monitor_* your email. You should receive an email with a subject *ALARM: "High throughput" in *. *_Review_* the email message. == Next section Click the button below to go to the next section. image::tear-down-workshop.jpg[link=../08-tear-down-workshop-environment/, align="left",width=420]