fiogf49gjkf0d Thanks for your response Raoul, I have tried but the 21 seconds were still there.
We monitored IIS with lots of tools (DebugDiag 1.1 to make a dump of the w3p.exe, Network Monitor to see the network trafic initiated by the w3p.exe), and we discovered that the IIS SLX process (w3p.exe) calls an external URL in order to make a survey (http://www.slxsurvey.com/?ver=7.5.3&per=60&lang=fr-BE) and that this website was deactivated.
This was not the main cause: actually our w3p.exe couldn't contact the URL because the w3p.exe was enable to detect our proxy. Sage should absolutly take this pbl into account by giving the possibility to the administrator to disable the survey. We deactivated teh sruvey that by changing datas in the DB. The procedure is described below (automatic translation from French).
After completing the loading, the w3p.exe DB will write to clarify that an attempt has already conducted survey date for the user in question.
A new attempt will be executed during the day. This explains why the only CRM was slow at first boot everyday.
The first time a user logs in, the CRM will specify that the next attempt DB shall be done 2 months later (and not the next day).
This explains why this problem appeared suddenly in mid-February, two months after the start of production.
The 21 seconds had no impact on CPU servers or bandwidth.
Workaround:
We have tested and then applied a PLSQL script in production that specifies for each user in the DB that the next survey will be attempted in 10 years.
Thus, the process no longer launches the IIS process of survey and 21 seconds away: So you go from 25 seconds to 4 seconds to the homepage.
It is also possible to add the proxy name in the web.config:
<system.net> <defaultProxy> <proxy usesystemdefault="false" proxyaddress="http://vsscansafe1e:8080" bypassonlocal="true"
/> </defaultProxy> </system.net>
|