View Issue Details

IDProjectCategoryView StatusLast Update
0005596SUMoRefactoringpublic2019-07-29 18:59
Reporterwolf Assigned ToKyle_Katarn  
PrioritynormalSeverityminorReproducibilityalways
Status acknowledgedResolutionopen 
Product Version5.9.4 
Target VersionShort term 
Summary0005596: Optimize SUMo startup
DescriptionIt seems that I still don't understand SUMo startup. What is SUMo doing at its startup so that it takes so long on a powerful non-gaming and no-workstation notebook. SUMo logfile reports needing 30 minutes on this powerful device. The only feature I enabled for startup handling was to check if there exists an update for SUMo itself. And I didn't deactive registry scanning. And I activated deep scanning. But this deep scanning feature shouldn't affect SUMo startup, just its scanning action which I didn't enable for SUMo startup.

I disabled for SUMo startup
* loading of SUMo on startup of Windows
* scanning for new software
* checking of updates for reported tool list
* minimization of SUMo into Info section
* beta version reporting
* reload of report list befor checking
* debug mode
Additional InformationWhen I read the standard log file after startup, I see some standard information reported. Including license check, it takes ONE second, including DB list. Does this mean DB list is loaded within a second?

Do I interpret it right that SUMo is performing a light scan on startup?
Do I interpret it right that this light scanning happens and overlaps with license checking?
How may I know when light scanning is finished?
Why does this light scanning happen at all?
Which kind of SUMo configuration disables such light scanning to see if this disabling turns SUMo startup in the operating mode I'm looking for?
What is SUMo doing when it reports loading in standard log?
This loading takes almost all these 30 minutes (minus 12 seconds)
During that loading phase, I read of removing, filtering and again removing. Is this part of loading or of overlapping light scanning?
It sounds part of (light) scanning.

My report list is quite large. My ignore list is larger then most SUMo users report list.
I can live with outdated data on SUMo startup. I can live with empty, incomplete or outdated update field on SUMo startup. When I need this field complete and up to date, I'll request a SUMo check action.
So are there already enough configuration options to achieve what I want?
What is wrong with my configuration to get the desired behaviour on SUMo startup?
Tagsstartup

Activities

Kyle_Katarn

2019-07-27 16:16

administrator   ~0003627

Last edited: 2019-07-29 18:59

View 2 revisions

Sorry but I don't have any particular clue on how to speed up the loading process, in particular with such "exceptionnally huge" list of software, which might exponentially decrease performance.
There is an overall improvement on large lists which I have to do (without being sure of being able to bring a significant improvement :-( )

Regarding startup, it takes about 2 secs on my "regular" PC with 50 items loaded (before automatic check, which can be disabled from settings)

wolf

2019-07-27 19:07

reporter   ~0003633

Will I get answers to my questions in the additional informations field, regardless if inside this issue or by email as you prefer?

I wrote what I want and consider acceptable. So at least for me, it seems that some startup steps are not needed and the effect of corresponding option is reaching less far then expected. Shouldn't there be different kinds of loading dependant on needs and usage?

So allowing to skip some steps (optional configuration possible), indexing and provide a light loading variant might improve performance for large data sets. And I've never ran out of memory regardless what I did on my computer. I doubled memory size by replacing memory of another device. And if needed, I already have the memory available but not yet installed to double it again. And I just received another SSD with better performance and size but didn't yet replace it.

And why should size impact performance decrease exponentially instead of almost linearily?

I don't know how you proceed various loading steps and how you use databases locally. Does your local database solution support indexing and do you load your data of local databases sequentially or in larger trunks?

And why have you set the state of this issue to feedback without writing what kind of feedback you're looking for?

Kyle_Katarn

2019-07-29 18:58

administrator   ~0003643

I fully concur to most of you analysis. Currently, the lead time for the startup operation is re-reading all exe file properties in order to extract accurate data (name,company,version) in particular in order to catch up changes made since last SUMo being closed.

One good way to improve this could be to use a local database and to reload only if file timestamp or date changed.

Issue History

Date Modified Username Field Change
2019-07-25 19:06 wolf New Issue
2019-07-25 19:06 wolf Tag Attached: startup
2019-07-27 16:16 Kyle_Katarn Assigned To => Kyle_Katarn
2019-07-27 16:16 Kyle_Katarn Status new => feedback
2019-07-27 16:16 Kyle_Katarn Note Added: 0003627
2019-07-27 19:07 wolf Note Added: 0003633
2019-07-29 18:58 Kyle_Katarn Note Added: 0003643
2019-07-29 18:59 Kyle_Katarn Note Edited: 0003627 View Revisions
2019-07-29 18:59 Kyle_Katarn Status feedback => acknowledged
2019-07-29 18:59 Kyle_Katarn Target Version => Short term