Forum Discussion
faziz_252121
Nimbostratus
NimbostratusLTM Round-robin
Hi
I have simple deployment, BIG-IP with two servers, but the problem is all traffic go to one server, and when disable that server it goes to second server although I enable round-robin;
why the traffic not distributed to both servers equally ?
@This is simple deployment I did not configured something like profile or iRules..etc
Thanks
@Faziz, You should press Control + F5, not F5 alone.
Pressing F5 alone will enable browser-cached results, even if the content on page has changed. The combination of Control + F5 will discard browser-cache.
nitassEmployee
I have simple deployment, BIG-IP with two servers, but the problem is all traffic go to one server
how much traffic did you test? do you know whether they all were hitting same or different tmm?
sol14358: Overview of Clustered Multiprocessing (11.3.0 and later)
https://support.f5.com/kb/en-us/solutions/public/14000/300/sol14358.htmlyou may try to disable cmp and see if it makes any difference (but cmp should be enabled in production). if nothing is changed, can you post the configuration here?
tmsh list ltm virtual (virtual server name)
faziz_252121I tried to do more Tshoot, and found when open browser inPrivate -no cache- and browsing vServer IP goes to server 1, then close the page and open again inPrivate and browsing vServer ip it goest to server 2, it works; I think this related to cache, but why I have to close the browser each time or delete history, bcz if i click F5 for refresh it stuck with one server I have to delete cache each time and this is not practical, i see some videos online they just click f5 for refresh and traffic swap between both server without delete cache- Ganesh_Gargdid you enabled persistence on VS??
- faziz_252121No, there is no presistence profile, and each time i have to delete history before refresh the browser, why? Anyone have experiance please
nitass_89166Noctilucent
I have simple deployment, BIG-IP with two servers, but the problem is all traffic go to one server
how much traffic did you test? do you know whether they all were hitting same or different tmm?
sol14358: Overview of Clustered Multiprocessing (11.3.0 and later)
https://support.f5.com/kb/en-us/solutions/public/14000/300/sol14358.htmlyou may try to disable cmp and see if it makes any difference (but cmp should be enabled in production). if nothing is changed, can you post the configuration here?
tmsh list ltm virtual (virtual server name)- faziz_252121I tried to do more Tshoot, and found when open browser inPrivate -no cache- and browsing vServer IP goes to server 1, then close the page and open again inPrivate and browsing vServer ip it goest to server 2, it works; I think this related to cache, but why I have to close the browser each time or delete history, bcz if i click F5 for refresh it stuck with one server I have to delete cache each time and this is not practical, i see some videos online they just click f5 for refresh and traffic swap between both server without delete cache
- Ganesh_Gargdid you enabled persistence on VS??
- faziz_252121No, there is no presistence profile, and each time i have to delete history before refresh the browser, why? Anyone have experiance please
- Hannes_Rapp_162
Nacreous
@Faziz, You should press Control + F5, not F5 alone.
Pressing F5 alone will enable browser-cached results, even if the content on page has changed. The combination of Control + F5 will discard browser-cache.
- faziz_252121@Hannes , this is works on Firefox when press control+f5, but with other browser doesn't, so i can confirm the issue related to browser
- Hannes_Rapp_162I did a brief testing, and it seems the exact implications of Ctrl + F5 vary across browsers (this may even vary across versions of the same browser). By standard, every web-browser has to omit browser-cache, and seems like all major browsers comply with that, however, on top of that, the Firefox browser opens a new TCP connection as well. That probably is the reason why you're re-balanced when using Firefox, and not in other browsers. As was already pointed out before, the balancing decision by default is made per unique connection that comes in from the clientside. Unless a 'Connection: Close' header is sent from the end-server, that connection (between Browser and BigIP) will be re-used for multiple GET requests. There are a number of possible circumstances that may contribute to different outcomes (i.e. Idle Timeout exceeded). So at least it works as you want in Firefox with the default configuration. With the use of custom configuration, there probably are ways to make Chrome and other browsers behave in a similar way.
- faziz_252121@ Hannes Rapp, thank you for your time(:
Hannes_Rapp@Faziz, You should press Control + F5, not F5 alone.
Pressing F5 alone will enable browser-cached results, even if the content on page has changed. The combination of Control + F5 will discard browser-cache.
- faziz_252121@Hannes , this is works on Firefox when press control+f5, but with other browser doesn't, so i can confirm the issue related to browser
- Hannes_RappI did a brief testing, and it seems the exact implications of Ctrl + F5 vary across browsers (this may even vary across versions of the same browser). By standard, every web-browser has to omit browser-cache, and seems like all major browsers comply with that, however, on top of that, the Firefox browser opens a new TCP connection as well. That probably is the reason why you're re-balanced when using Firefox, and not in other browsers. As was already pointed out before, the balancing decision by default is made per unique connection that comes in from the clientside. Unless a 'Connection: Close' header is sent from the end-server, that connection (between Browser and BigIP) will be re-used for multiple GET requests. There are a number of possible circumstances that may contribute to different outcomes (i.e. Idle Timeout exceeded). So at least it works as you want in Firefox with the default configuration. With the use of custom configuration, there probably are ways to make Chrome and other browsers behave in a similar way.
- faziz_252121@ Hannes Rapp, thank you for your time(: