Perhaps this could be part of a collaboration project where we users get to design and build the tool?
A kind of open-source thingy.
We could even extend the functionality of ControlCenter. Add a button "Analyze logs" which will send the log to some WS and a user gets a link to a webpage with results after processing. This will improve awareness of streamshare resource amongst our user database...
It could be great if you can track and/or correlate StreamStudio, Service Gateway and StreamServers logs from diferent enviroments at the same time.
This can give you a global picture of what is going on.
Great idea Vyv, once confidentiality of customer information isn't breached i'm all for this idea.
Great idea!
This has my thumbs-up.
I would even suggest a hibride kind of analiser.1: Online, to make log analisize2: Offline, for local use.
As a matter of fact: I'm working on a 'dashboard' that can read logs and shows status of applications realtime. Complete with alarms and all. This might be compattible.
So, yes.... Thumbs-up. :-)
Here are 2 "Use Cases" that may be helpful to scope the tool. Any tips on these would be welcome.
Use Case 1:Dan works as a system administrator for a host of applications, every day he checks the smooth running of about 5 applications and their related hardware, most often on virtual server systems. He has Microsoft & RedHat training and experience. StreamServe has been implemented within the last year and now all of the implementation team are gone - leaving the StreamServe system running on auto. Dan has very little training in StreamServe - in fact he has just attended one workshop where a StreamServe Implementor has described the important components and features as well as how to administer the system. The Implementor is available via mail / phone - yet charges a minimum of 3 hours for every call / incident raised by Dan's company. If a StreamServe related problem or a question should turn up from the Business Team or the IT Department then Dan feels reluctant to call the Implementor directly as the cost can outweigh the issue. Help would be at hand for Dan if he uses a log file analysis tool - here he could get a collected report with useful links and direct solutions to known issues.
Use Case 2:
Marie is a Sales Administrator - She is in charge of the Billing Process for all Sales at a National company. Amongst many concerns and areas of responsibility - her concerns about StreamServe include making sure that invoices are not sent twice, that all emails have been sent and received (no bounces), that the print shop has received the daily invoicing file and that all e-invoices have been set OK. A daily / weekly run of log file analysis will help her identify any issues for any customers / partners.
I have been playing with the idea of a collaborative log message database on StreamShare since we first started to plan for StreamShare. My idea seems to go hand in hand with Vyv's so rather than posting an alternative solution of a competing idea, I will try to explain it here:
From the log in CC, I wan't to be able to click on a message that I wan't to know more about. A log message is basically just one sentence, and it can be hard to explain the full meaning of an error, warning or info message in just one sentence. The user can then click on the message in the log and he will be taken to a page on StreamShare. This page is specific to the log message (based on log id). On this page, I can find a longer explanation of the log message, possible causes and effects, links to related log messages. It is in this context also possible to list related forum posts, blogs and articles.
There are literally thousands of log messages in the OpenText StreamServe apps, and I don't expect it to be possible to fully document each log message as described above without a significant effort. To overcome this, i suggest that we add a community section to each message page, where our users can comment on the log messages. This will allow us to find log messages that can be improved or explained better more easilly. It also allows the community to help each other by providing further examples, possible causes for certain messages and so forth. The community section will not be a forum where discussions can be held.
This would be like a mix between Oracles ora code db and Microsofts MSDN Community Content
It should of course be possible to search for a particular log ID on StreamShare and be taken to the correct log message page.
Hi Stefan,
When developing and running several tests – this log link function that you suggest here would be very useful indeed. Normally in these situations the developer looks closely at the log file for errors, warnings, information to get an appreciation for how their solution is working. This step is performed repeatedly. Log files are most likely small, and when developing, the entire flow of a test can be reached from the Control Center.
Yet we should tailor any enhanced ‘log analysis’ solution to include business users and administrators less familiar with the inner workings of StreamServe where log files are huge and may even be wrapped over several days. Business users and seasonal administrators want to analyze log files for any errors, duplicates, warning messages and so on over a period of PROD/LIVE operation. Any solution should therefore encompass this need. Thus a function to scan a log file and produce a list of results and suitable actions should therefore be also included.
I have already built one that produces the following results as a result of a customer requirement:
A summary:
LOGFILE SAP\log_file.txtVERSION StreamServe Communication Server 4.1.2 SP1 Build 1071LICFILE License file:C:\Program Files\StreamServe\4.1.2\Server\strs.licFILES-PROCESSED 365ODBC USED YESODBC CONNECTION ERRORS 40FATAL ERROR COUNT 11DRIVER ERRORS YESJOB COMPLETE ERRORS 87
And then all ERRORS:
Timestamp Line # Log ID Message0711-155456 2013 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0711-155746 2085 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0711-160327 2160 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0711-160337 2291 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0711-160344 2363 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0711-160418 2438 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0714-154847 13302 2660 exported_configurations\Runtime.dux(1): Fatal error: syntax error0714-155631 13503 3465 Fatal error: No resources allocated for process, possible reason: CallProc() not called during preproc phase0815-150822 14387 1233 drivers\PCL5e600_AXIS_1.0.drs(3): Fatal error: Driver already defined:HP PCL5e 6000821-155146 15347 1236 drivers\axis_win.drs(27): drv WIN: Fatal error: Option already defined:PaperSource0821-155706 15353 1233 drivers\win.drs(5): Fatal error: Driver already defined:WIN0901-093317 26545 1346 PDF [Image] error: image load failed - logo01.bmp0901-093317 26550 1346 PDF [Image] error: image load failed - logo01.bmp1015-151111 30569 1346 PDF [Image] error: image load failed - logo01.bmp1015-151111 30573 1346 PDF [Image] error: image load failed - logo01.bmp1019-101717 31453 1076 XMLIN: XML parse error (message, line, column, position): "encoding specified in XML declaration is incorrect", 1, 30, 30
So you can get a feel for the type of errors that commonly occur and group them by frequency / type and so on.
Even the above report is quite technical - yet the customer has improved this since I saw it last - to include feedback from Status Messenger and I beleve that a business user scans this once a week.
Ah, So what you are really looking for is statistics reporting?
Point number 1 in my proposed solution covers statistics for a Business User use case, point number 2 would be used for Business users / Admins to hone in on issues where error codes that you propose could also be presented (along with other links to ancilary posts / support). It should summarise these issues and allow users to find interesting points without having to scan large log files manually.
So a web based statistical report generator with dynamically allocated search results based on warnings / errors.
In my last post I mentioned a tool that already does this - I would like to post it onto this IDEA - yet cannot find any attachment functionality. I shall try to post an article with the tool and link it here. It needs to be cleaned up a little first.
Thanks Vyv,
Then I think I understand. Attachments are not allowed on ideas, so just post it in the forum and link to it from here.
About five years ago I wrote a Streamserve log file analyzer to analyze errors and processing times. It was capable to analyze version 3 and 4 Streamserve. I wrote it in AWK, a scripting language that's very easy to learn. If it can help you, I'd be pleased to share it.
Hans Bloemendaal, the Netherlands
