JazzMon Server Monitor User Manual
Contents
1. ecm common ISc mService acceptC Place chart ombined aptinternal zervicd 3 rect PlanRiestSer la As new sheet Chart 3eponse Time z vice getPlanSeare hRecults2 e sem common ISc mService updateC Hac As object in service et vg cms me Gens th ce Let Second catego Second value Y axis e Step 3 Type in a Chart Title and axis labels if desired o In some cases the data may stack up vertically like towers If this occurs check the Axis tab in Step 3 and change Category x axis to use Category instead of Automatic e Step 4 Set Chart Location as new sheet worksheet in spreadsheet Hit Finish filesystem common IFilesystemService getFil Average Reponse Time eTreeByVersionable3 8 scm common IScmService getHistoryForVer sionable scm common IScmService acceptCombined 3500 apt internal service rest IPlanRestService get PlanSearchResults2 scm common IScmService updateCompone nts2 9 process internal service web IProcessW ebUI Service getContributors 1 filesystem common IFilesystemService comp areWorkspaces scm common IScmService batchCommit 3000 2500 2000 scm common IScmService suspend 1500 filesystem common IFilesystemService getBl ameWS scm common IScmService resume 1000 scm common IScmService discardChangeS ets s2. The CCM or jazz application is generally where most of the traffic occurs If you are still looking at the jts workbook follow the hyperlink back to the top level work book and then select the jazz subdirectory workbook from there The first two charts provide an intermediate scale summary of the traffic frequency and time by totaling web service traffic by the major server subsystem components This gives you a break down by the different functional areas of the server In an SCM site it is typical for the scm component to show significant traffic because it involves file downloads The build and repository components often display a regular pattern of activity in support of the build engines checking for and processing build requests A list of component names is included in the table of contents for each workbook JazzMonDemo Analysis JazzMonDemo Analysis jazzdev torolab ibm com jazz xls Table of Contents Chart Tab Source Data Tab Description Component Counts servicecomponents etCnt Server Components Counts Component Total Time servicecomponents etTot Server Components Total Processing Time Service Avg Resp Time service etAvg WebServices Average Response Time Service Counts service etCnt WebServices Counts Service Total Time service etTot WebServices Total Time Service Downloads service bsTot WebServices MegaBytes Downloaded Service Uploads service brTot WebServices MegaBytes Uploaded Async Avg Resp Time async etAvg AsyncTasks Average
3. sem common IScmServi 120000 EEEE ee RA 2008 E gettstorForViroi 100000 al IQueryRepozitorySere able e count 500 ecm common IScmServi 1000 ceacceptCombined filezystem common File ContentService PUT aptinternal zervice rect l PlanReztService getPlan Searchiezultz2 sem common SemServi ce updateComponentz2 k 3 d 3 d repocitory common inter nal RepozitoryRemotes ervice fatchOrRefrechlte Vlad Mad Wied Iard Wied EFI DER vedete 53 88 17 146 175 20 23 Series workitem common ini 4 workitem common int workitem common inia workitem service inte J 7i ada move z Counter etCnt trend_etCnt 1 1 Category X axis labels HN scm common Some scm common IScmSe apt internal service r scm common IScmSe 7 Name Values Benove Category X axis labels service etAvg B 1 C5 1 service etAvgl A 2 Tania nga SE Copyright IBM Corp 2012 17 of 32 Chart Wizard Step 3 of 4 Chart Options Titles Axes Gridlines Legend Data Labels Data Table Chart title lverage Reponse Time Average Reponse filesystem commo nJFilezystemServi ce getFileTreeBy y ersionableS 9 scm common ISc mService getHisto a aiass Chart Wizard MN Chart Location l te e Category X axis Value Y axis
4. software JazzMon Server Monitor User Manual Version 1 4 0 December 17 2012 IBM Rational Performance Engineering Team Dave Schlegel aMrnn 3 IBN Supported B Eo sii EL E 3 1 2 Download ird abrir P 3 eEeruunail m 3 LA Server Tuning References ete tre p oni tre E erat teet etes eet oie irre eter tints 3 2 Data CollectiOn Processine aT E elm 4 2 1 Running JazZMO ME 4 22 Command Monitor semina aue n n a A E E a aie a E 4 2 3 Command Gather crate eur TET vos tas cava e T Ea 5 24 Command AnalyZ6es iiec fal isos deve RU on HR o Pete ot SEP eM REP UR ERE 5 24 1 Clustet Node Data AgbBregation cec her tera e Eee redi bcp eer ae 6 PRISE B M 7 2 0 Command Password ett entities LEE IEEE es Rv shal oe EEA ea FUIS Ve Eee PU etes Ve ro NAE 7 2 7 Command Version oet aet ehe TE PO cee aa VIS D va eee Yen Eve In seeders va Uses e Pee den Vea ee Ve oer Pene Geass 7 NMPV AuSvnnrrr 8 EB Jose auteni i qe V 8 Se lL Whatis a Web Servic i 8 31 2 WED SELVICES e seteeoe ret eH UD Dee E
5. Copyright IBM Corp 2012 5 4 2 Counts service_etCnt The Service Counts chart provides insight into what the most frequently called operations are Be aware that web services called many times may not consume the most processing time if their average response times are small See the next section for total time a better indicator of load Here are a few highlights 600 000 500 000 Items build internal common ITeamBuildService getB uildEngine f build_internal common ITeamBuildRequestSemi 400 000 ce getNextRequest repository common internal IRepositoryRemote Service fetchOrRefreshltems repository common internal IFeedService GET 300 000 Wn NN build internal common ITeamBuildService save 100 000 Calls Eg N S scm common IVersionedContentService GET downloads a single file This is usually the most frequent call made in a conventional SCM environment and performance issues here can have a major impact on overall performance The file size for this call does impact response time so look at the Bytes Sent data service bsTot csv to see how your average size compares to the baseline size before drawing overall conclusions build internal common ITeamBuildService getBuildEngine and ITeamBuildRequestService getNextRequest are two calls that are made frequently in support of the build engines waiting for build requests repository common internal IRepositoryRemoteService fetchOrRefreshItems is
6. IMemberPhotoService GET 0 009 a o7 0 065 0 003 35 4496 E M 4 gt Mservice etAvg m gt Ready Sum 109580 45 Copyright IBM Corp 2012 21 of 32 4 3 2 Sorting Time trend tables often have over 500 rows one for each of the available Jazz web services To make sense out of this wealth of information we must extract the most important items by sorting and filtering In Excel select the entire table by selecting the upper left corner of the table use ctrl A then select Data Sort from the menu bar For working with averages select Avg and sort in descending order to bring the longest response times to the top of the table To return to the original alphabetic order select Counter etAvg Ascending Sort Sort by CENEEKEEKENES 2e lt ending Descending Then by Ascending Descending Then by Ascending Descending My data range has Header row D No header row aos C Microsoft Excel service etAvg csv 2 lea ial File Edit View Insert Format Tools Data Window Help Type a question for help x f Counter etAvg A cs ier CU CV cw E Counter etAvg jazzjazz 1 20 2012 9 46 Totals Max Avg Avg Baseline T lirlesystem common IFilesystemSerice getFileTreeBy Versionable3 3 395 79 368 6674 2984 127 70 257 2106 66 El scm common IScmService getHistoryForVersionable 0 207 5 074 5 532 2 393 716 58 234 28132 39 3 E scm common ISc
7. n Chart Reponse Time service etAvg L m i li gt ze i Ze CUN Be Ba ie es ee Bh Draws I AutoShapes 10 amp j z 8 L3 Ready The resulting data table may have too wide a data range to chart easily but it will highlight the average times of the most frequently used operations This particular table shows some unusually slow operations in the most frequently used operations Copyright IBM Corp 2012 20 of 32 4 3 Working with Data Tables and Charts The charts produced using the JazzMon_Visualizer focus on the top N operations as sorted by the Totals column You may want to focus on other web services by sorting by other values such as averages or actual numbers in a particular time slot etc By resorting or filtering a table you can investigate other web services the associated chart will be updated as soon as the table changes 4 3 1 Basic Table Structure This provides a spreadsheet showing counter names and average response times above e Column is the counter being measured with the leading com ibm team leader trimmed off e Column 2 is the optional baseline that the collected data is being compared to based on ANALYSIS BASELINE e Intermediate columns are the data snapshots If there are too many columns to read in the full spreadsheet use the ANALYSIS SAMPLE TIME property to reduce the sampling rate and run analyze again e The last 4 columns provide built in
8. Category X axis Crossesat 0 Response Time Seconds Display units None z Show display units label on chart E Logarithmic scale 7 Values in reverse order Category X axis crosses at maximum value The resulting chart will scale up the smaller values of primary interest From the adjust chart shown below we can see that compared to the baseline file average times a number of operations spike dramatically mid day under load Keep in mind that baseline data just provides an average reference point for an alternate server or point in time But unusually long response times that correspond to user reported performance problems usually provide a starting point in identifying what web services are involved in performance issues Looking at the average response times is interesting but without some additional context can be misleading e Average response times are calculated based on total time and counts during that interval for the service This naturally makes the graph somewhat spiky because the values return to 0 when there is no traffic in a given period e Some operations normally take a substantial amount of time especially reports and integration operations e Discard Suspend and Resume can take significant time depending on size of the end users suspended queue Copyright IBM Corp 2012 28 of 32 Average Response Time Top 20 sorted by Totals WebServices spuoses ow 1 esuodsay 29 of 32
9. Response Time Async Counts async etCnt AsyncTasks Counts Async Total Time async etTot AsyncTasks Total Processing Time JazzMonDemo Analysis Server Components Counts Top 20 sorted by Totals 600 000 e scm _ 1 repository build 500 000 workitem process 400 000 a ny 200 000 ill UNE 1 100 000 4 LLL RUP UP GU Counts enar Mi pitt reine Copyright IBM Corp 2012 27 of 32 5 4 Web Service Traffic Details 5 4 1 Average Response Time service_etAvg Average Response Time is an important indicator of end user experience The patterns you are usually looking for are changes in response time or unexpectedly long response times NOTE If you find the default Y scale is hiding the details you need mouse right over the values in the left hand Y axis and adjust the scale to a specific volume as shown below You may also hide rows containing specific web services in the data table to filter them out the chart will reflect the change immediately JazzMonDemo_Analysis WebServices Average Response Time Top 20 sorted by Totals interop common service lInteropService getE xt ernalManagerTypelnfo z reports common internal service IReportRestSe rvice postRenderQuery 2 000 social common internal ISocialRestService ge Value Y axis scale Auto V Minimum 0 7 Maximum 250 V Major unit 500 V Minor unit 100 V
10. Wn PT LUE RET e M LR EL LS eas T PT jazzdev torolab ibm com qm Calls 100 000 oS aS Ns 4 Cn yw ow The indicates a missing data sample Po o Se gh mh E hg nd quay SA MM xv x See Section 4 4 for more information QN ade M NY ON LD N nh SKS S V VS ANS SN XN CAR SH NHN SI NEN SEN FREE PO LEE Qr QE ANON GN ANY AN MU GANE AN v NIRE NON ue g Copyright IBM Corp 2012 25 of 32 5 2 Floating License Usage license_flVal in jts application To find out the primary usage cycles click on the jts subdirectory workbook link The Floating License data tab will provide a clear indication of user activity highlighting the types of user licenses checked out and active during the main work period for your users Matching user cycles to other charts like average response times will usually highlight which operations are under performing during the periods of greatest interest JazzMon Analysis Floating Licenses CheckedOut Top 20 sorted by Totals Oteam rtc stakeholder token Oteam rrc contributor token Oteam rrc contributor floating Ei rtc developer iep token m rqm contributor token m rqm contributor floating Counts f QO QN AW Dd aS Ar AA ON Qr SN SG SAM Praha alae at AON wx y j VS QV M Me NY M M ne ON ade cade VY ade cae cade NT ade cad cade RN ade cade VY ade cade VEY ade CSS SIGN SESS e n Copyright IBM Corp 2012 26 of 32 5 3 Component Summaries
11. Z System specific operations e dashboard Support for web browser dashboard presentations showing mix of reports and queries e datawarehouse Services unique to managing meta data about the repository e enterprise Enterprise extensions e filesystem Manage versioned file artifacts between local workspaces and repository e fulltext Full text search capabilities e interop SCM integrations between Jazz and external systems including work item synchronization with ClearQuest e jfs Jazz Foundation Services Resource based storage and query services providing access to the JFS repository and user information also used by JFS based fronting applications e links Access and manage links between different types of artifacts such as work items and change sets e process Process definition controlling activity flow roles and permissions with customization e reports Reports provide data about activities and artifacts over time in the repository e repository User license and server administration services along with modeled storage services for persistent and query Copyright IBM Corp 2012 10 of 32 e rtc Rational Team Concert one lone operation e scm Source Code Management basic change set management operations check ins accept deliver suspend discard resume workspace management e social Support for Open Social integration e vs Visual Studio client platform support e workitem Work item opera
12. at the bottom of the client to view and manage the data The resulting report will record the web service traffic generated by this individual client in performing whatever operation you do Copyright IBM Corp 2012 8 of 32 Jazz Metronome https jazzdev torolab ibm com 944 3 jazz DAR Services Item Manager Connection Service Method Count Time s Time Avg s Worst s IRepositoryRemoteService 1 515 46 0 379 0 75 fetchOrRefreshItems 1 515 46 0 379 0 75 l2 IScmService 0 813 0 163 0 609 batchCommit 0 609 0 609 0 609 getChangeSetLinkSummaries 0 204 0 051 0 063 0 11 0 11 0 11 0 109 0 109 0 109 0 343 0 343 0 343 0 392 0 131 0 157 0 392 0 131 0 157 getComponentStateSummaries interpretChanges compareWorkspaces 2 IFileContentService storeContent 4 4 5 1 4 IFilesystemService 3 0 562 0 187 0 343 1 1 1 3 3 Call Count 15 Elapsed Time 3282 s Item Count 1839 Cache Size 3599477 B For example checking in a few files produced the output above The web service counter names are related to the names shown above i e the fetchOrRefreshItems web service has a full name of com ibm team repository common internal IRepository RemoteService fetchOrRefreshItems Using Metronome you can relate which user operations call which web services For the web browser client using a product like Firebug will let you see the direct traffic as well Keep in mind that the web service reports show how
13. formulas to compute information about the data columns that provide sort keys to highlight the more interesting data NOTE If you get an error in the Avg Baseline column see more information on ANALYSIS TARGET to adjust the data for the spreadsheet application you are using o Totals Max Avg sum maximum average of all data columns in the row respectively o Avg Baseline compares Average to the Baseline value for the current row as a percentage e NOTE The etAvg data tables now have two additional columns o TotalTime column is added to show the total time spent on each web service o TotalCounts column shows counts for the run and can be used to filter out or hide low count web services o Average of the averages is computed as TotalTime TotalCounts to provide a weighted average for the run Eile Edit View Insert Format Tools Data Window Help fe Avg Baseline cs CT 1 20 2012 9 46 Totals Avg Baseline 1 Counter etAvg jazz jazz apt common resource ResourcePlanningSerice checkWritePermissions 0 069 0 1 0249 0 016 22 5390 apt common resource IResourcePlanningService fetchContributorlnfo 0 523 0 376 217 188617 0 219 41 83 apt common resource IResourcePlanningService findVVorkLocation 0 005 0 004 2 1 326 0 018 369 89 apt internal common rcp lIterationPlanService checkPermission 0 024 0 o 0102 0 001 4 4796 apt internal common rcp llt
14. i eiie du mede de nte rers deett 9 3 1 3 Web Service Components M R B w 10 SE ESCIDNIMIFMET TH B 11 31 9 Floating License WU Sage 11 5 1 6 Distributed Object Grid Cache iue ete t eee petes etr AE Ea iru bns IRE R i 12 3 2 Reposttory Reports ioi HERR aE A HIERRO HR HEBEL INE CH dgavesegsGuecscsseunag ONDE PETERE HERREN AAEREN HERR CHE ARET 12 9 9 SELVES ITO cesserit iR ORDEI need M mute en 13 3 4 State Cache Counter Repott sessie loe eee eei rele ed Oe per eH terit cel repens ir ese oa solere Pepe 13 4 Vis alizing Data HR 14 4 1 Charting using the JazzMon Visualizer cicer neneeese aiaei raa EE E EE Ei easi 14 42 Charting Manually 5 rU t ERR UE ONUS ERN EE E A ARAA 16 42 1 Re ding CSV Elles RATER ORE APR EE ER AME Ie IRE 16 LEAD D 16 4 2 3 Combining data sets optional seiten een esba rt ye ausnavecotevsyecevseeuscadasuyecuuarabecotevetersssvoensodavagevtasraseceteets 19 4 3 Working with Data Tables and Charts nennen nnnm ENEE re tenete sense EEE ERR 21 4 3 L Basic Table Structure uer rete EEXXVE E e SERES E INEEEE E E SV A Er ee PERVE Se SERERE ENS EF e VERRE e EINER RE NERA NN danse 21 4 32 TIL 22 4 SS Nul e EE 23 4 4 Re
15. rest IPlanRestService getPlanSearchResults2 2 135 of 434r 1990 235 981 1295 cm common IScmService suspend 12 564 iu 0 689 270 57 74 I3 scm common IScmService batchCommit 0 288 0 0 56 760 572 2683 91 A filesystem common IFilesystemService getFileTreeByVersionable3 3 335 79368 667472984 2106 66 346 scm common IScmService acceptCombined 4 601 a 16 448 3890 548g 890 0496 357 process internal service web IProcessVVebUlService getContributors 12 302 a 3 056 934 3124 79 94 M 4 gt M Chart Reponse Time service etAvg Drawy AutoShapes Aa OON Filter Mode NOTE Since the _etAvg data tables include the TotalTime and TotalCounts you may want to first sort the table in descending order by TotalCounts and hide or filter out the low frequency calls i e hide anything below 10 calls per hour in order to ignore slow response web services that are not called often enough to be interesting 4 4 Reporting Gaps during monitoring When the server fails to respond to a request for a web service report JazzMon will make a copy of the previous report as a place holder and leave a file CounterContentServerX ERROR txt that contains the exception This may result when the server has gone down or is being rebooted or a communications error occurs temporarily JazzMon will continue to make requests for subsequent reports for the remainder of the run Placeholders are used to ensure that data samples across multiple
16. size_ave_per_state average size per state o size ave per item average state size per item not very useful orm size size taken by all the ORM Object Relational Mapping tables of the item type o orm size prct total percentage of total ORM size o orm size prct namespace percentage of just this namespace s total ORM size o orm size ave per item average ORM size per item content size size taken by all the content associated with items of this type o content size prct total percentage of total content size o content size prct namespace percentage of just this namespace s total content size o content size ave per state average content size per state not very useful o content size ave per item average content size per item stored content size actual persisted compressed content size taken up by all content associated with items of this type o stored content prct total percentage of total stored content size stored content prct namespace percentage of just this namespace s total stored content size stored content ave per state average stored content size per state not very useful stored content ave per item average stored content size per item stored content ave compression ratio of stored content to content size lower value means higher compression a e E e 3 3 Server Info The server information report captures basic information about the server uptime maximum memory total memory Java VM Jazz build etc It is a snap
17. wie S qv wie Soh oo S c Se aa Shh Mat de See a ys SA g gY ON V eV Ss gC IU ES xv AEP WY FPP Pe of PoP oF ye vO Oh M e S C CHEE S SS S C CX CK ON e IC aS SR S we CELL qv These operations are used for things like Cleaning up obsolete data results BuildResultPrunerTask Notifying users and processes of completed operations BuildNotificationTask notification mail workitem service save postnotification Supporting build engine operations BuildSchedulerTask BuildAgentLoop Taking snapshots for data warehouse operations ScmSnapshotTask CommonSnapshotTask NOTE Some web services like reports are not run very often and can have very long response times Baseline data comparisons are also vulnerable to large spikes in response times which may be skewing averages Sometimes data points that suggest a serious response time regression can be attributed to a single large sample i e a user running a huge report Using counts median max and standard deviation metrics can help identify when these issues affect average response time accuracy Copyright IBM Corp 2012 32 of 32
18. 42 100 00 2 95 100 00 5 7996 10 5396 led gt nn R to N N 0000r2r IN oW UO Sao XXX 43 M 4 gt YIN Chartl service etcnt i Drawy J AutoShapes more Ready e Go past the last column of service_etAvg csv and paste the two columns past the Avg Baseline column You may need to use mouse right Paste Special and select Values to paste in the results of the formula based column from service_etCnt csv Paste all Validation Formulas All except borders i Column widths Formats Formulas and number Formats Comments Values and number Formats Operation None Multiply Add Divide Subtract Skip blanks Transpose Copyright IBM Corp 2012 19 of 32 jazz jazz filesystem common IFilesystemService getFileTreeByVersionable3 3 335 scm common IScmService getHistoryForVersionable 0 207 apt common resource IResourcePlanningService checkVVritePermissions 0 069 apt common resource IResourcePlanningService fetchContributorlnfo 0 523 apt common resource IResourcePlanningService find VorkLocation 0 005 apt internal common rcp lterationPlanService checkPermission 0 024 apt internal common rcp lterationPlanService fetchPersonalPlan g 4 M Chart Reponse Time service_etAyg Is i in e Scroll down and spot check that the rows are using the same names to make sure the data is align
19. 5 5 Asynchronous Tasks async_etTot The total time table for asynchronous tasks is also interesting to check to see how much time is being spent on background tasks for maintenance and supporting operations These operations can be long running so just because an operation finished in a given interval it doesn t mean it was only running during that interval But if you find that long operations are completing during or shortly after a period when performance problems have been identified it may indicate that some maintenance operations are taking longer than expected and not finishing in the off hours period Elapsed Time Seconds 20 000 16 000 iie E ad 14 000 BuildResultPrunerTask 12 000 updateLinksTask 10 000 x versionedContentCleanup 8 000 6 000 4 000 2 000 JazzMonDemo_Analysis AsyncTasks Total Processing Time Top 20 sorted by Totals e RunBuildCommandTask 18 000 a Defect140238Re NM BuildAgentLoopla Snaps Runner ask OutgoingSyncTask ExternalProxySyncTask notification mail DeleteditemScrubTask LicenseRemoteSnapshotTask ProcessChangeEvents workitem service save postnotification BuildSchedulerTask EN UE pta Bh de du T LLL MUI S B Ei ortum BC meae c ean sae can A X P MAP 0 d aO a a9 a9 a a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a9 a a9 a9 a9 a8 Pr y a arca Pra a ee
20. Symphony to filter and visualize the data to put the reports in context The next chapters provide more information on how to work with the analysis output This step is automatically performed when monitoring is completed unless analysis in place is disabled You can use the analyze command to see intermediate results while monitoring is still running or to reanalyze the data if it wasn t completed for any reason The analyze command properties are set to reasonable default values but can be adjusted as appropriate e ANALYSIS_DATADIR is baseline data location by default the Data directory within the JazzMon installation e ANALYSIS_BASELINE is the name of a baseline set within ANALYSIS_DATADIR If you don t want any baseline comparison set ANALYSIS BASELINE to nothing ANALYSIS BASELINE The baseline set has one or more files providing baselines for one or more server types ccm jts etc The default is a weekday Jazz net baseline e ANALYSIS SAMPLE TIME allows analyze to skip intermediate data samples if desired For example if you collect hourly data for 30 days you may only want to see the data on a daily basis set ANALYSIS SAMPLE TIME to 1440 minutes 24 60 NOTE This parameter is also applied to adjusting the baseline data to the right proportions e ANALYSIS AGGREGATE LIST provides a list of application suffixes that guide how to aggregate cluster node data together into a cluster wide report Clusters should be monitored by monito
21. adar em Surface 9 Bubble uon 9 sem common SemServi ce getHiztoryForVerzioi able sem common ISemServi ceacceptCombined aptinternal zervice rect l EN em z SearchRiezultz2 2 scm common ISemServi ce updateComponents2 Data range service etAvgl A 1 CS 20 Series in Columns ine with markers displayed at each data alue Press and Hold to View Sample Cancel Back Einish Step 1 In the Chart Wizard select Chart Type Line and then hit Next Step 2 Reduce the data range to the top 20 web services by changing the last number in the Data Range i e CW 320 becomes CS 20 Given the wide value scale later rows may not show on the chart and earlier versions of Excel have a 255 column limit Then select Series in value of Rows and hit Next o Sometimes the X axis at the bottom shows numbers instead of dates times To fix this select the Series tab in Step2 If the first row Counter lt field gt shows up in the Series list remove it and select the first row to be the Category x axis labels as shown below 80000 60000 40000 20000 0 1 30 Wizard Step Source D E Data Range filezystem common IFile 9 sem common ersioned al 160000 ContentService GET mE zystemService getFileTr 4 3000 sByVerzionable 140000 7 2500 9
22. ch as logging in checking in a change set updating a work item or downloading a file The best way to understand what web services do and how they are used is to enable the Metronome feature on the RTC Eclipse client which tracks and reports on what web services have been executed by the individual client For more information see https jazz net blog index php 2008 02 01 the jazz metronome tool keeps us honest Preferences Jazz Source Control General Ant Context Aware Search Maximum number of threads 10 Help Install Update Instant Messaging Java This action will repair metadata associated with files and Folders loaded from Plug in Development a repository workspace Existing file content on disk will not be changed however the repair may cause differences between your eclipse and Run Debug i z repository workspaces to show up as changes in cases where the metadata is corrupt Please review in the Pending Changes view all your changes after fH C S running this action to make sure they are intended Content Transfer Repair Metadata Team deae Ignored Resources EIFE rli tit rz Jazz Source Control Models Metronome Team Process Work Items Show traffic statistics in the status bar Restore Defaults Apply You can enable the Metronome feature by visiting the Window Preferences user interface and selecting Show traffic Statistics as shown above and then use the Metronome icon
23. cm common IScmService deliverCombined 500 scm common IScmService updateCompone nts scm common IScmService createWorkspac e d filesystem common IFilesystemService getFil 4 ul eTreeByVersionable4 M3 reports service rest internal service Reporta KS d bleService GET 7 scm common IScmService accept This produces a chart of the average response time over the 4 day period that looks like the chart above By hovering over individual lines you can query which web service and what time slice you are looking at Copyright IBM Corp 2012 18 of 32 4 2 3 Combining data sets optional Sometimes you may want to combine the data from different tables perhaps looking at average response times sorted by counts so you can look at the average times of those operations called most frequently One way to do this is to cut and paste the average counts column from service_etCnt csv and paste it into the average response time spreadsheet e Unhide all the rows in both spreadsheets and sort both by the Counter lt field gt so the rows are exactly in the same order control A to select all then Data Sort by the first column in ascending order e Select the first column counter name and then control click on the Avg column from the service_etCnt csv file CT CU jazzjazz Totals Max j Awg Baseline Fr 1m 10 0096 370 1058 r 9 7496 SHE 20 6296 1 1 0596 2 27 89 3
24. dService getBuildEngine is in the build component of the server Copyright IBM Corp 2012 9 of 32 e Service totals trend tables these appear in the top level directory to compare the total traffic across all the servers being monitored in the run serviceTotals etCnt csv etc 3 com ibm team repository service internal counters contentservice CounterContentService Mozilla Firefox DEAR File Edit View History Bookmarks Tools Help a com ibm team repository service internal co i wre xA web service comter 1 dapsel time seconds avg or ira debug team commen service Team Dehug ence ad1User 0002 ern ihaa debug team common service Team Dehug erice getDehug essions By amp tiribortes 0004 comibm debug team common service ITeam Dehug ence addUser 0001 com bra debug team common service ITear Debug vice is Installel 0002 orn ibra debug team commen sevice Team DehugS ence setUser ChenfVersionf 0001 lt i 3 1 3 Web Service Components This is a short summary of what the different system components com ibm team lt component Name gt lt service gt do e apt Agile Planning and Tracking build Support for Build Engines to access and process build requests e calm Collaborative Application Lifecycle Management C ALM specific operations e com ibm debug team internal debugging e com ibm teami I System specific operations e com ibm teamz
25. e properties and either cut and paste out snippets you need or make a full copy to use as your jm properties file So for example let s say you use a different property file name jmTest1 properties and want to provide your password from the command line to keep it separate it s only needed by the monitor command then you would type this java jar JazzMon jar monitor file jmTest1 properties SEQ PASSWORD myPass java jar JazzMon jar gather file jmTest1 properties java jar JazzMon jar analyze file jmTest1 properties 2 2 Command Monitor The monitor command collects web service counter reports and optionally repository reports from one or more server applications or hosts see chapter 3 for more information The reports are collected in a run output directory under separate Copyright IBM Corp 2012 4 of 32 subdirectories for each URL The output directory defaults to c temp JazzMonRuntime Windows or var tmp JazzMonRuntime Linux but can be modified by changing the PATH OUTPUT DIR property This allows you to monitor different sets of servers simultaneously or to keep different runs separate You will be asked for permission before existing output is overwritten Unless analysis in place is disabled JazzMon will automatically perform the analyze step when monitoring is complete This command requires the following properties to be set e SERVER URL LIST is a comma separated list no spaces of one or more server URL s
26. e same folder based on the OutputDirName allowing you to zip up all the workbooks to facilitate sharing with other team members e NOTE Pressing the Create Workbooks button will automatically close any conflicting spreadsheet files with the same name that are already open Make sure to save any work you want to save or to change the OutputFileName or OutputDirName parameters to avoid overwriting work you want to keep Once you have the results you want skip to section 4 3 for more information for how to interact with the data E JazzMonAnalysis xls l ees E B JazzMon Visualyzer JazzMon Analysis Table of Contents Chart Tab Source Data Tab Description Total Service Counts serviceTotals_etCnt Total WebServices Counts Total Service Total Time sericeTotals etTot Total WebServices Total Time Total Service Downloads serviceTotals bsTot Total WebServices MegaBytes Downloaded Total Service Uploads serviceTotals_brlot Total WebServices MegaBytes Uploaded Total Async Counts asyncTotals etCnt Total AsyncTasks Counts Total Async Total Time asyncTotals etTot Total AsyncTasks Total Processing Time Subdirectory workbooks JazzMonAnalysis jazzdev torolab ibm com jazz xls JazzMonAnalysis jazzdev torolab ibm com jts xls JazzMonAnalysis jazzdev torolab ibm com qm xls Notes 1 Services are Web Services sent from clients to server to perform specific operations 2 Asynchronous Tasks are background tasks the server p
27. e substantially longer to produce than web service snapshots In some larger repositories these may take a half hour or more and they should be run less frequently Consider having a separate JazzMon run that is taking these perhaps once a day or once a week namespace states states_prct_total items items_prct_total com ibm team applicationmigration 0 0 0 0 0 0 com ibm team apt 4 0 0 4 0 0 com ibm team apt plansnapshot 0 0 0 0 0 0 com ibm team apt resource 0 0 0 0 0 0 com ibm team apt snapshot 0 0 0 0 0 0 com ibm team build 182503 12 0 180947 15 7 com ibm team compatibilitypack 212 0 0 108 0 0 com ibm team dashboard 0 0 0 0 0 0 com ibm team diagnostictests 13 0 0 13 0 0 repoReport lt n gt txt e namespace functional area within the overall repository e states total number of item states for this namespace changes o states prct total namespace states percentage of total states e items total number of distinct items for this namespace o items prct total percentage of total items Oo ave states per items average states per item e size size taken by all this namespaces states excluding content o size prct total percentage of total size o size ave per state average size per state o size ave per item average state size per item e orm size size taken by all the ORM Object Relational Mapping tables for this namespace o orm size prct total percentage of total ORM size o orm size ave per item average ORM size per item e content size
28. ed correctly then you can delete the Counter etCnt column and rename the service_etCnt csv Avg field as Avg Counts e Sort in descending order by the new sort column Counter et amp vg jazz jazz Awg Baseline AwgCnt Counter etAvg jazz jazz Avg Baseline AvgCnt 2 workitem serice internal oslc IOSL CService GET D 029 4955 77 95 35 706 74 E E scm common lVersionedContentService GET 0 006 1171 05 3 978 83 Ea repository common internal IRepositoryRemoteService fetchHandlesByLocation 0 089 i 770 8896 2437 72 5 repository common service IQueryService queryDatalnContext 0 186 1065 6596 2324 21 8 repository common internal IContributorRestService postContributor 0 56 437 9896 2307 41 7 build internal common ITeamBuildService getBuildDefinitionStatusRecordsForCont 0 586 17 4095 2 285 73 apt viewlets internal service IPlansSerice getProjectAreaByTeam 0 001 25 26 1 823 63 9 repository common jauth ICheckToken GET 0 001 281 05 1 430 36 10 repository service compatibility internal IJtsConfigurationStateRestSerice getJtsCi 0 100 00 1 415 85 11 workitem coramon internal WorkltemRepository Service fetchNewer 0 007 478 0596 514 07 12 repository common internal IRepositoryRemoteSerice fetchStates 0 133 19 1095 369 22 13 workitem common internal rest IQueryRestService postGetResultSet 0 188 775 7896 279 74 14 workitem common internal rest IQueryRestService m 0 005 405 A 263 57 M 4
29. erationPlanService fetchPersonalPlan 1 043 1 2847 205 62 775 3 106 297 8096 apt internal common rcp llterationPlanService fetchPlanProgress2 0 796 o 3 0974 0 036 4 50 apt internal common rcp llterationPlanService fetchPlannedWorkltems 0 691 or 67 30723 0 701 101 4095 apt internal common rcp llterationPlanService fetchResolvedlterationPlan2 0 899 or 24 8578 0 251 27 8996 D apt internal common rcp IlterationPlanService fetchResolvedWorkltem 0 003 0 o 0005 0 000 3 8596 apt internal common rcp llterationPlanService fetchWorkltemProgress 0 109 of 40 6 982 0 423 387 66 apt internal common rcp llterationPlanService save2 0 236 o 2 0 586 0 019 8 2396 apt internal common teamload ITeamLoadSerice fetchTeamLoadlnformation 0 158 0 1 04257 0 015 9 55 apt internal common wiki WikiPageAttachmentSerice GET 0 01 0 o 0 000 0 000 0 00 apt internal common wiki WikiService allowsEmbeddedxhtml 0 001 07 o 00057 0 000 12 63 apt internal common wiki WikiService createPageUsingOwner 0 052 0 o 0000 0 000 0 00 apt internal common wiki WikiService findAttachments 0 018 o o 01607 0 004 21 81 apt internal common wiki WikiService findPageUsingOwner 0 008 or o 0 0317 0 001 10 66 9 apt internal common wiki WikiService saveAttachment 0 07 or o 0 000 0 000 0 00 D apt internal service rest LinkRestService postGetLinks 0 979 2782 266 35 327 2 798 285 78 apt internal service rest
30. erforms for maintenance or as side effects of Web Services 3 JazzMon is a monitoring tool providing insight into server traffic see http jazz net library article 822 Component Descriptions apt Agile Planning and Tracking build Support for Build Engines to access and process build requests calm Collaborative Application Lifecycle Management C ALM specific operations E eam ihm dahun tanm internal dahupnina Y M 4 gt M TableofContents Total Service Counts serviceTotals etCnt lt m r Copyright IBM Corp 2012 15 of 32 4 2 Charting Manually 4 2 1 Reading CSV Files The best files to start with depends on what performance issues are already known but these are good starting points e service etTot csv helps identify which web services take the most total elapsed time and represents count multiplied by average e service etCnt csv helps identify volume of traffic counts to find the most frequently called operations e service etAvg csv isolates the average response times for web services computed to show the average per analysis sample time interval totalTimePerInterval countsPerInteral instead of relying on the server s original running average The interval average helps identify poor performance at specific times of day In Excel e File Open navigate to the file c temp JazzMonData run0 lt host gt service_etAvg csv You will need to change Files of type to select Text Files pr
31. es a standalone Java jar executable documentation release notes and sample baseline data 1 3 Getting Help The JazzMon software is provided by the Rational Performance Engineering Team Please ask support questions on the Jazz net forums https jazz net forum using the tag jazzmon 1 4 Server Tuning References JazzMon helps to visualize actual server performance but doesn t tell you how to improve performance For more information on how to properly tune a Jazz server see the following articles e Tuning the Rational Team Concert 4 0 server https jazz net library article 1029 e Collaborative Lifecycle Management 2012 Sizing Guide https jazz net library article 814 Copyright IBM Corp 2012 3 of 32 2 Data Collection Processing To get started using JazzMon right away see the new JazzMonQuickStart one page guide in the install directory and refer back to this manual as needed 2 1 Running JazzMon JazzMon is a runnable jar file that you run from the command line java jar JazzMon jar You must have a version of Java already installed to launch the jar file If needed you can use the version of Java that is packaged with the RTC Client under jazz client eclipse jdk jre bin for JazzMon for RTC 3 0 and above use Java 1 6 or later for JazzMon for 2 0 0 2 use Java 1 5 or above Use java version to check your java version Without arguments it provides a basic help message java jar JazzMon jar co
32. for this spreadsheet to enable Buttons m 3 Select location of JazzMon Analyze output FolderName lt path gt Source of JazzMon Data after Analyze command run Browse for FolderName Either t in directly or use button to browse l 4 Adjust output properties MainTitle JazzMon Analysis Main title included in all charts TopValues 20 How many of the top data rows to include in chart OutputFileName JazzMonAnalysis Base name for output files will have xls appended S OutputDirName JazzMonAnalysis_Spreadsheets Absolute path or relative subdirectory unt lt FolderName gt ProcessSubDirs TRUE Process application or server subdirectories creating workbook in each IncludeBaseline TRUE TRUE to include baseline column B FALSE to not show it StartDataColumn Optional Blank or name of starting data column by name i e AA to show subset EndDataColumn Optional Blank or name of ending data column by name i e AZ to show subset TimeLabelSpacing Optional Blank or number of columns between time axis X axis labels 5 Create interlinked Excel workbooks of data tables and charts Create Workbooks M 4 gt nh Summary la uD r Follow the instructions on the Summary page 1 Run Jazzmon to monitor and analyze your data See the JazzMonQuickStart one page Read in JazzMon Visualizer xls and enable macros You will either be prompted for whether to enable macros or not or you may see a banner at the top telling you macros are disabled
33. mService acceptCombined 4 601 16 448 3890 548 889 40 951 890 0496 ISl apt internal service rest IPlanRestService getPlanSearchResults2 2 135 4341 1990 235786 20 947 981 1295 Hsc common ISemSeriice updateComponents2 5 161 20 302 1806 488 330 19 013 368 3996 ME process internal service web IProcessVVebUlService getContributors 12 302 3 056 934 312481 9 834 79 94 dm ilesystem common IFilesystemService compareVVorkspaces 1 08 2 703 885 101 622 9 318 862 8196 scm common IScmService batchCommit 0 298 0 56 760 572398 7 998 26839196 12 564 0 689 270 733 7 255 57 74 2 68 0 646 490 338 6 802 253 7996 Pl scm common IScmSerice resume 12 98 0 555 424353 5 843 45 02 El scm common ISemSenice discardChangeSets 2 317 1034 533 145210 5 615 242 3496 1 51 2 023 481 38 678 5 066 335 51 19 085 0 430 95 136 4 529 23 7396 Slllscm common ISemSeice createWorkspace 5 281 O 428 163583 4 507 85 34 fAlfilesystem common FilesystemService getFileTreeByVersionable4 3 697 2 669 406 90 052 4 276 115 657 2 146 Dr 384 303786 4 046 188 5496 1741 0 765 362 173004 3 813 219 02 20 2 996 o 300 256510 3 156 105 3496 FAM apt internal common rcp llterationPlanService fetchPersonalPlan 1 043 1 294 295 62 775 3 106 297 8095 M4 gt serie et vg jj f e m gt i Draw jj AutoShapes Nw OO 4 d i DA Or Ae Ready Sum 3977086 924 O Copyright IBM Corp 2012 22 of 32 4 3 3 Filtering Fil
34. mmand file propertyFile lt property gt lt value gt where command is one of the following monitor Start monitoring one or more servers gather Collect copy of data into permanent location analyze Analyze data to produce time trend tables for visualization baseline Create new baseline from a pair of monitor snapshots password Prompt for password and show obfuscated equivalent version Display JazzMon version The basic order of operations is to run monitor for some period analyze the data to generate trend CSV files and then visualize the CSV files as Excel or Symphony charts and tables NOTE By default JazzMon will perform analysis in place monitoring to get the data then automatically analyzing the data when finished if this is not enabled you will need to run the gather command in between monitor and analyze You control what the command does by providing a set of properties that identify what server s to monitor provide login information how long to run and so forth These properties are provided in a property file but can also be set from the command line when appropriate By default JazzMon looks in the local directory for jm properties other locations can be specified using the file command line option A simple default version of jm properties is provided and requires only the URL s of the server s you want to monitor and your login information If you need to use more advanced properties see jmTemplat
35. mon internal IRepositoryRest Service GET repository common jauth llssueAuthToken P OST build internal common TeamBuildService se tLastContactTime repository service IVersionCompatibility Rest Service GET repository common internal IRepositoryRem oteService describe repository common internal IFeedService GE T 15000 10000 repository service IItemRenderService GET 5000 workitem common internal WorkltemReposi toryService fetchNewer build internal common ITeamBuildService st artBuildActivity Feb 28 2012 Feb 28 2012 Feb 28 2012 Feb 28 2012 Feb 28 2012 orkiem a i aa 17 10 17 11 17 12 17 13 17 14 ee EM build internal common ITeamBuildService ad NOTE One thing that stands out in many charts is the sudden dip in the second column for all web services This is an artifact of data computing data values in time intervals In order to get the relative number of counts or time the analyzer compares each time slice to the slice before it to get a delta count so the counts for the first time slice are 0 s for the second slice is Time2 Timel etc This first column dip artifact can be eliminated by hiding that first time column in Excel by selecting the first data column and using mouse right Hide as shown below Copyright IBM Corp 2012 24 of 32 5 Interpreting Results 5 1 Overall Totals Top Level workbook JazzMon produces a set of CSV files in the to
36. much time the server required to perform the operation Metronome data also includes the round trip time between the client and server If the server time is relatively small the difference between the two may represent excessive network latency which may be the real cause of perceived performance problems Itis also important to realize that some web services are used extensively by multiple components of the system and don t just support specific use cases i e com ibm team repository common service IQueryService queryItems is used by many operations 3 1 2 Web Services Each web service provides three groups of data values covering elapsed time et bytes sent or downloaded to clients bs and bytes received or uploaded from clients br JazzMon produces a series of time trends extracted from this table shown below computed for each server URL being monitored e Service trend tables time trend tables by individual web services for each interval o Service etAvg csv has elapsed time averages in seconds o Service etCnt csv has the number of times counts that a web service is called o service etTot csv has the total elapsed time spent per web service in seconds etCnt etAvg etTot o Service bsTot csv bytes sent totals o service brTot csv bytes received totals e Component trends time trend tables aggregated by system component based on the web service name i e com ibm team build internal common ITeamBuil
37. n txt csv to see the csv files produced by JazzMon e Adjust the column widths for readability select all control A and then double click the bar between the A and B column headings to auto size all columns e Split the window pane to make it easier to work with large spreadsheets o grab the little rectangle at the right end of the horizontal scroll bar and drag it between the B and C columns o grab the rectangle at the top of the vertical scroll bar and drag it between rows 1 and 2 o scroll the right pane to see the right end of the rows You can optionally format formula results for better readability select each column click mouse right and Format Cells e Totals works best as Number format thousands separator and no decimal places e Maxand Avg work best as Number format thousands separator and 3 decimal places e Avg Baseline works best as Percentage format 2 decimal places 4 2 2 Basic Charting Using this sorted spreadsheet you can create a chart using the Chart Wizard e Select the entire data set by once again clicking the corner box e From the main menu click on the Chart Wizard looks like a column chart Copyright IBM Corp 2012 16 of 32 Standard Types Chart type Data Range Custom Types filezyztem common lFile zyztemService getFileTrq sByVerzionableS 3500 Chart sub type 3000 lad Column E Bar we XY Scatter Doughnut er R
38. n larger repositories these reports may run 30 minutes or more 2 3 Command Gather This step is not needed as long as analysis in place is enabled The gather command copies the collected data from the monitoring output directory to a more permanent location which will also contain the analysis data generated in the next step There are no required properties for gather By default it will copy the data from the PATH OUTPUT DIR to temp JazzMonData runO If you want to save the data in a better location provide the following properties e ROOT NETWORK SHARED DIR LINUX is the main storage path for Linux platforms The default is lt temp gt JazzMonData e ROOT_NETWORK_SHARED_DIR_WINDOWS is the main storage path for Windows platforms e RUN ID is the subdirectory within the destination area by default run0 Gathering data provides a snapshot of the data for the analyze command to produce time trend tables from You can run gather at any time while monitor is running or after it is completed 2 4 Command Analyze The analyze command post processes the data to produce a series of time trend charts focusing on individual variables in the web service counter reports number of operations average response time etc over time It creates comma separated text files Copyright IBM Corp 2012 5 of 32 csv where each report snapshot is a column in the table These reports can then be manipulated in Microsoft Excel or Lotus
39. nd tables that can be used to investigate and visualize how well a server is performing Note JazzMon supports the use of baselines to provide a way to compare newly gathered data against an earlier time period or a different server site to provide some context for data interpretation They put the new data in context to help differentiate what seems normal versus interestingly different but must be used with caution there is no right set of numbers that all servers will match all the time Response times and activity vary based on many variables time of day number of users other activities etc Comparing performance between sites can be interesting but misleading comparing current performance against an earlier baseline from the same server can be much more meaningful This manual describes how to install and use the JazzMon package providing a brief overview of captured data and illustrating how to interpret and visualize this data to gain insights into Jazz server performance 1 1 Supported Platforms Operating Systems Windows Server 2008 2003 Windows XP Windows 7 Red Hat Enterprise Linux 5 x Debian Rational Products Compatible with Rational Team Concert and other Jazz based products at version 2 0 1 GA and above 1 2 Download Site Installation JazzMon is available as a zip archive at https jazz net wiki bin view Main JazzMon Download and unzip the archive to a local working directory The package includ
40. p level folder that provide data on overall totals across the set of application or server URLs you are tracking The initial workbook that the JazzMon Visualizer produces looks at this information to provide a 50 000 foot view of the traffic across these servers Note The examples in this chapter are based on a week s worth of data collected from the main jazzdev production server Table of Contents Chart Tab Source Data Tab Description Total Service Counts serviceTotals etCnt Total WebServices Counts Total Service Total Time serviceTotals etTot Total WebServices Total Time Total Service Downloads serviceTotals bsTot Total WebServices MegaBytes Downloaded Total Service Uploads serviceTotals brTot Total WebServices MegaBytes Uploaded Total Async Counts asyncTotals etCnt Total AsyncTasks Counts Total Async Total Time asyncTotals etTot Total AsyncTasks Total Processing Time Reviewing the data in this workbook provides an overview of the comparative traffic across your applications or servers Typically the JTS application experiences low volumes of traffic since it is usually providing user authentication and services in support of the primary applications like ccm or jazz or qm These charts often highlight major patterns JazzMonDemo Analysis Total WebServices Counts Top 20 sorted by Totals 800 000 700 000 e gt n 9 jazzdev torolab ibm com jts 500 000 Ea a ECSPBESNE i ATTE DICE fel
41. porting Gaps during monitoring sseesesseesseeseeeeene eene nennen nent nnen nennen retne nnne tnit eret retener entree nee n nne 23 5 irauossruradt dre R 25 5 1 Overall Totals Top Level workboOk 4 1 eripit ote treten eodeni eee reine PR eee rhe a n nep ee yx EE e Ee PEE snsvaraieusedveses 25 5 2 Floating License Usage license flVal in jts application nennen nennen rem rennen 26 3 3 Component Summari ES M 2f Copyright IBM Corp 2012 1 of 32 5 4 Web Service Traffic Details 5 4 1 Average Response Time service_etAvg 5 4 2 Counts service etCnt 5 4 3 Total Time service etTot 5 5 Asynchronous Tasks async_etTot Cop yright IBM Corp 2012 2 of 32 1 Introduction The JazzMon server monitor package collects gathers and analyzes Jazz Server performance data allowing for analysis of trends over time and comparison between separate monitor sessions runs For an overview see JazzMon Seeing what your server is up to https jazz net library article 822 To get started using JazzMon right away see the new JazzMonQuickStart one page guide in the install directory and refer back to this manual as needed JazzMon provides a runnable Java jar file that collects performance snapshots from one or more Jazz servers over a period of time then post processes the data to produce time tre
42. rently used The lt H gt value is used to divide the baseline by the number of hours to get an hourly average Example In the default 8 hour baseline jazz ccm 200users S8hrs txt the site name is jazz i e jazz net the application is ccm and then after the double underbar __ it specifies 200 users and 8 hours When a new baseline is created the initial filenames contain the last part of the server URL that was being monitored as part of the name The lt name gt part can be edited but the app pattern is needed i e jts ccm etc to identify which baseline should be used for data being compared to it If there is a mismatch between the app name generated from one set of source data and the app suffix in target data being analyzed later make adjustments i e if you make an RTC SCM baseline from one server with an app suffix of ccm but want to compare it to another RTC SCM server using the jazz suffix just make another copy of the file containing ccm in the name with jazz instead To use the new baseline set copy the new output directory to your ANALYSIS DATADIR directory if needed and modify the ANALYSIS BASELINE property to use the new baseline set name for subsequent analyses 2 6 Command Password The password command will prompt the user for their password then print out an obfuscated version of the password that can be used in the properties file Password entry masking is not supported due to c
43. ring each application on their individual nodes then enabling aggregation allows true cluster wide reports to be computed e ANALYSIS AGGREGATE ZERO BASIS enables a mode that deducts the initial web services report counts and totals before calculating output in order to simulate restarting the server e ANALYSIS CLUSTER enables generating trend and total tables from Distributed Object Grid data when available e ANALYSIS EURO LOCALE enables reversing comma and periods from locals using commas as decimal points e ANALYSIS TARGET default Excel specifies the anticipated spreadsheet application that will read in the data to adjust formulas included in the output For IBM Lotus Symphony specify Symphony no quotes case insensitive e ANALYSIS IN PLACE default true If enabled automatically analyze data in original monitor output directory without need to gather 2 4 1 Cluster Node Data Aggregation When JazzMon is used to monitor a server cluster it needs to monitor the individual nodes separately and then aggregate combine the data from the nodes to get an accurate picture of the overall cluster Using web service reports from the load balancer front end used for most operations will collect a random jumble of reports from the individual nodes that are not meaningful in most situations For example this is how you would monitor two applications on a two node cluster SERVER URL LISTzbluesws01 torolab ibm com 9443 jazz blueswsO01
44. s for the counts licence flVal csv in the JTS application directory output but does not aggregate the information at this time Copyright IBM Corp 2012 11 of 32 3 1 6 Distributed Object Grid Cache When Jazz 4 0 products are clustered an additional report may be produced that provides information about the Object Grid communications traffic used to synchronize information between nodes JazzMon creates time trend tables for attempted count objectgrid_atCnt csv elapsed successful count objectgrid_etCnt csv elapsed average time objectgrid_etAvg csv and elapsed total time objectgrid_etTot csv This information is best reviewed in consultation with IBM support 3 2 Repository Reports Repository reports use an internal API to collect data about the contents of the Jazz repository itself providing insight into the distribution of different types of artifacts in the repository based either on the component level repoReport x txt or in the more detailed report by individual types of artifacts in the component namespace repoReport itemized txt The columns of greatest interest are the number of items unique artifacts and states changes to those artifacts Other columns show what percentage a namespace is compared to the overall total or additional information about storage size Shown below is a small segment of the overall table NOTE These reports require the user have Jazz Administrator access to the repository and tak
45. servers or nodes all have the same number of samples to avoid data skew because of gaps The corresponding data column is highlighted with to indicate it is not valid data and the data columns may be hidden if desired The placeholder data is a copy of the previous sample which in this simulated communication error causes an apparent jump in the data between 17 11 and 17 12 is identical so the computed delta count drops to zero but then jumps dramatically Rebooting the server will reset all counters to zero so a server restart may cause a similar spike in some reports before settling down again and in some cases may become a negative value Hiding the placeholder column and following column will provide a more meaningful representation NOTE Aggregated node data makes some basic assumptions in order to merge data from multiple nodes See the section 2 4 1 Cluster Node Data Aggregation for more details Copyright IBM Corp 2012 23 of 32 scm common VersionedContentService GE Counts showing error T 8 repository common internal IRepositoryRem oteService fetchOrRefreshltems build internal common ITeamBuildService sa ve 3 repository common service IQueryService qu eryltems X build internal common ITeamBuildService ge tBuildEngine 9 build internal common ITeamBuildRequestS ervice getNextRequest t jfs users service IContributorService GET 30000 25000 20000 repository com
46. shot of the contents of this URL https lt yourhost gt 9443 jazz service com ibm team repository service internal IServerStatusRestService ServerInfo 3 4 State Cache Counter Report The State Cache Counter report provides information about internal cache management traffic Copyright IBM Corp 2012 13 of 32 4 Visualizing Data Each time trend file displays one variable from the web service counter reports as a series of columns as the value changes over time e To work with the data in Microsoft Excel use the JazzMon Visualizer macro file in the installation directory e To work with the data in Lotus Symphony read the comma separated CSV files as spreadsheets for filtering visualization and analysis Note If using Lotus Symphony adjust ANALYSIS TARGET property to avoid formula error ANALYSIS_TARGET symphony These examples demonstrate Excel using a data set created by monitoring and analyzing the target server for four days 4 1 Charting using the JazzMon_Visualizer The JazzMon_Visualizer is an Excel workbook that provides macros to automatically read in most of the analyzer CSV files and turn them in to charts automatically B JezMon Visualizerds terea JazzMon Data Visualizer Version 1 4 0 December 4 2012 Step Property Value Description 1 Run JazzMon to monitor gather and analyze web service reports See https jazz net library article 822 for basic information 2 Enable macros
47. size taken by all the content associated with this namespace s items o content size prct total percentage of total content size o content size ave per state average content size per state not very useful Copyright IBM Corp 2012 12 of 32 o content size ave per item average content size per item stored content size actual compressed content size taken up by all content associated with this namespace s items o stored content size prct total percentage of total stored content size o stored content size ave per state average stored content size per state not very useful o stored content size ave per item average stored content size per item o stored content size ave compression ratio of stored content to content size lower value higher compression repoReport lt n gt itemized txt namespace item specific item type within a given namespace states total number of states of this item type o states prct total percentage of total states o states prct namespace percentage of just this namespace s total states items total number of distinct instances of this item type o items prct total percentage of total items o items prct namespace percentage of just this namespace s total items o ave states per items average states per item size size taken by all the states of this item type excluding content o size prct total percentage of total size o size prct namespace percentage of just this namespace s total size O
48. tering can be done by manually hiding data rows or columns select rows or columns and mouse right hide or using the Filter AutoFilter feature of Excel under the Data menu Either approach will immediately affects any charts based on that data table Select all the data control A then select Data Filter AutoFilter This will add pull down menus under every column that allow you to select Top 10 or any Top N to automatically hide rows not selected by the filter selection To undo a given selection use the filter pull down menu to select All again The autofilter Custom option will allow you use AND OR logic to specify ranges and additional filtering of strings with SQL like contains and equals expressions ix Microsoft Excel service etAvg xls i8 File Edit View Insert Format Tools Data Window Help UJ ig TT 1006 i Arial o C3199 Y f 4 341 Counter etAvg jazz j 1 20 2012 9 v Totals Max ounter etAvg v jazz ja Ly Totals Max C A 1 20 2012 8 1 20 2012 9 v Total M Fr 3 filesystem common FilesystemSerice getBlameWS 2 58 0 646 490 JIE 253 79 cm common ScmService getHistoryForversionable 0 207 0 5 074 5532 239 28132 39 ilesystem common lFilesystemService compareWorkspaces 1 08 a 2 703 885 101 8628196 cm common IScmSerice updateComponents2 5 161 20 302 1806 488 368 39 pt internal service
49. tions define create edit delete query work items 3 1 4 Asynchronous Tasks Asynchronous tasks are background processing tasks that the Jazz server carries out internally for maintenance and other processing not related to a specific user request JazzMon creates time trend tables for the counts async_etCnt csv elapsed time Count average response times async_etAvg csv standard deviation async etDev csv and total time spent async etTot csv A sample of this table is shown below Top level totals are also computed asyncTotals etCnt csv etc com ibm team repository service internal counters contentservice CounterContentsS BAR File Edit View History Bookmarks Tools Help com ibm team repository service internal co f e a ibm com https jazzdev torolab ibm com 9443 jazz service E asynchronous tasks CR T Le x mm 9 f em raza of ne raza Fee sezor aw ce a count n n eme eme e owa ie nexa 9 xw em oas ono ov weal o 7 4 nni n Dil AT mm Som DAT MIT lt TR 3 1 5 Floating License Usage The JTS application server may be tracking Floating License usage information recording when new floating licenses are checked out or when they expire This information provides an indication of how many users are active but does not show any users with permanent licenses so don t rely on it as an absolute indication of current usage JazzMon creates time trend table
50. to monitor You may monitor different server hosts or multiple application servers on the same server i e myhost 9443 ccm myhost 9443 jts e SEQ USERNAME is the user name for logging into the server s If the supplied user is an administrator on the target system it is also possible to monitor the size and growth of the repository by enabling repository reports with the RUN_REPO_REPORTS property e SEQ PASSWORD is the user s password If set to prompt then the user will be prompted each time default If provided in the properties file the password can be provided in either clear text or obfuscated form see Command Password section below It is up to the user to maintain password security Other properties that can be adjusted include the following e SEQ RUN LENGTH ARG default 7d controls how long the monitor will run Server data collected over a period of time will provide a better idea of the ebb and flow of traffic over a week Its values can either be the number of iterations to run i e 8 for 8 snapshots or a time duration such as 8h for 8 hours or 7d for 7 days e PARM COUNTER RATE MINS default 60 minute controls how often data samples are taken e RUN REPO REPORTS off by default enables gathering detailed repository content reports the login user must been an Administrator for these reports PARM REPOREPORT RATE MINS controls how often those reports are taken 480 minutes 8 hours by default i
51. torolab ibm com 9443 jts A bluesws02 torolab ibm com 9443 jazz bluesws02 torolab ibm com 9443 jts Aggregation is enabled using the ANALYSIS AGGREGATE LIST of application suffixes as follows ANALYSIS AGGREGATE LIST jazz jts When node data is aggregated reports from different nodes are merged to make a cluster wide report there are some adjustments made in order to combine the data e Averages are computed based on dividing total time between reports by the total counts between reports to get a weighted average for the overall cluster e Standard Deviations are NOT aggregated correctly at this time but until there is proper support for this the maximum of the standard deviations will be shown This value is taken from the node which had the highest standard deviation and that it is not suitable for statistical tests because it does not represent the standard deviation of the aggregated population of samples Copyright IBM Corp 2012 6 of 32 2 5 Command Baseline Baselines provide a way to for the analyze operation to compare data against an earlier time period or a different server site to provide some context for data interpretation They put the new data in context to help differentiate what seems normal versus interestingly different but must be used with caution there is no right set of numbers that all servers will match all the time Response times and activity vary based on many variables time of day number of
52. until you click a button to enable them Some sites may hide this option by default for their employees If you have trouble talk to your local administrator Copyright IBM Corp 2012 14 of 32 di Select the location of your data Press the Browse for FolderName button or type in the path to your analyzed data Select any CSV file or the servermonitor txt file to pick the directory Adjust the output properties These allow you to adjust the titles used in charts how many of the top N rows will be included select the output file name and folder locations select a sub range of data to chart and other chart options Press the Create Workbooks button to create a series of workbooks e The initial work book you see is the top level totals comparing overall traffic between the different servers applications or cluster nodes This will have hyperlinks to the individual servers or application workbooks e Fach server or application will have its own workbook with links back to the top level workbook to allow navigation across the data The name is based on the OutputFileName and the subdirectory name e neach work book the Table of Contents page provides links to the different data table tables tabs in the current spreadsheet The top left cell in each table provides a hyperlink back to the Table of Contents to assist in navigation For each data table tab the preceding tab is the corresponding chart e Workbooks are output to th
53. urrent Java limitations It is up to the user to maintain password security 2 7 Command Version The version command just displays the current JazzMon version number for reference with support issues or feature content Copyright IBM Corp 2012 7 of 32 3 Jazz Performance Data JazzMon collects data from one or more Jazz based servers by saving a combination of repeated snapshots of some reports web service counter reports optional repository reports and one time snapshots of others server overview and state cache report This chapter provides a basic description of the information being collected 3 1 Web Service Counter Reports Web service reports provide a wealth of information about historical traffic and performance information for the application server The basic report is available to any user by visiting the following URL on the target server https lt yourhost gt 9443 lt app gt service com ibm team repository service internal counters CounterContentService The port 9443 may vary There are two key tables in this report that JazzMon analyzes web services asynchronous tasks and two additional tables that it analyzes when they are available floating license usage and distributed object grid caching for clusters 3 1 1 What is a Web Service A web service is a low level individual request sent to the Jazz server Multiple web services are often needed to carry out an individual user operation su
54. used to obtain many individual elements of data used in work items and other RTC artifacts repository common internal IFeedService GET performs a feed query for a specific project area such as Build events Team information or workitem changes In the RTC Eclipse client go to the Team Dashboard view go to the Event Log and click on the down facing triangle to get to the Configure menu for feeds This will list the feeds a given client is watching using this web service if you edit an individual Feed source you can see how often that feed calls this web service to get updated information JazzMonDemo Analysis WebServices Counts Top 20 sorted by Totals e scm common lVersionedContentService GET s repository common service IQueryService quer T La WANN xr S ao Oa ao SPM ASA FOP oP gd S Copyright IBM Corp 2012 30 of 32 5 4 3 Total Time service_etTot The actual elapsed time spent processing the web services is a better indication of where the server is spending more effort than the raw service counts It helps put average times and counts in context showing by showing the product of the two Slow response times for operations that are called frequently can be a better indication of where problems may lie Elapsed Time Seconds JazzMonDemo_Analysis WebServices Total Time Top 20 sorted by Totals ly 20 000 3 15 000 10 000 5 000 5 Copyright IBM Corp 2012 31 of 32
55. users other activities etc But using baselines helps to isolate where performance or traffic are dramatically different and worth further investigation Each baseline file identifies what server it is from ccm jts etc and includes the sampling duration in the filename to allow hourly rates to be computed but also as a guide to what sort of period the data represents an 8 hour stretch during the daytime a 24 hour day showing night activity and maintenance jobs a full week including weekends etc The baseline command creates a new set of baseline files from monitor output data using the same source data location as the analyze command In addition it takes two properties e ANALYSIS DATA RANGE sstart end provides the suffix numbers of the data snapshots to use The baseline s will be the difference between these two snapshots 10 18 compares CounterContentServer10 html to CounterContentServer18 html and marks it as an 8 hour snapshot The comma separated list can t have any spaces e ANALYSIS OUTPUT DIR directory identifies where the baseline files should be written The default is the current ANALYSIS DATA DIR NewBaseline see Section 2 4 Each server subdirectory under the source data will be turned into a new baseline file with the server name and a user time suffix in the format name app N users H hrs txt If you know the number of users rename this filename to reflect that but the user count is not cur
Download Pdf Manuals
Related Search