GUPs problem
Hey,
to make an idea of the network: 1 SEPM (11 7101.1056), over 800 computers connected (500 over gigabit connection, rest over wan), database - embedded, server Dell Power Edge 2950 with HyperV host 2008 Standard - VM win 2003 x64 ( 4 cores + 4 GB ram given to SEPM ).
i could say so far so good with our configuration but like 2 - 3 months ago (before RU7 MP1 upgrade ) we noticed an increase in traffic over different moment of the day. thought that Ru7 MP1 will solve the issue so we upgraded, the problem still persists.
the locations with +5 computers in same LAN have each of them an active GUP configured to limit it`s traffic to 10 MB/ 5 MB / 256 kb depending on the location and our routers. The lowering of the bandwidth + increasing the time for direct connection in case GUP is not reachable dropped the issues over the network from 3h+ of max consume to 30-45minutes.
as the problem still persisted we found a nasty thing, let`s try a small configuration example
SEPM, 5 GUPs - G1, G2, G3, G4, G5 and each of them has 5 clients G11,G12, ... G15 ..., G21 ..and so on.
some facts :
- over the gigabit connections, all G11, G12 ... connects 100% of time to G1 and takes updates, while G1 connects to SEPM.
- over gigabit + VLANs, G21, G22 ... connects to G1, but now the funny things starts when G1 connects randomly to G2, G3, G4, G5 with no apparent reason, and most of time instead of using the fastest connection to SEPM, goes over WAN to another GUPs from 2 MB location
- over WANs i see other interesting things, G31, G32, G33 instead of connecting to their declared and active GUP G3, they connect to G1 from another WAN, most of the time all to the same G1 (random selected over the day ) and rarely to other GUPs ex. G21 to G1, G22 to G2 - as it should, G23 to G1, G24 to G3.
the only solution we could use was to remove the GUP policies over the locations where this nasty thing appears
so, ... any of you encounter something like this? is there any solution ?
Comments
What your actual
What your actual requirement??
Regard
AJ
a solution to make the GUPs
a solution to make the GUPs work as they should ?
if i set a GUP to a specific group of computers i expect those computers to connect only to that source of updates and later on to the console management if the GUP does not answer
and not
as it happenes now, with random connection to another computer on same VLANs which is not part of its group, and of course is not their GUP :(
Hi, Is all the GUP's In
Hi,
Is all the GUP's In the same subnet ???????
nope, different subnet we use
nope, different subnet
we use 255.255.255.224
interesting enough is that they go to the first ip or the first GUP from that ip range
"- over WANs i see other interesting things, G31, G32, G33 instead of connecting to their declared and active GUP G3, they connect to G1 from another WAN, most of the time all to the same G1"
in my case the most annoying group is for like 60 computers spread on groups with 6-7 each with their GUP and they connect to the 172.25.2.1 < which is in my example G1 :(
as i said before, strange is that on any client outside G1 i look everything appears good in settings/registry(gup setting checked their too) but when it comes to download updates, they randombly connect to G1 even if their GUPs is available (every GUP is a w2003 server 24/7 running 365days)
Kindly find the attach
Kindly find the attach link
Document on GUP creation
https://www-secure.symantec.com/connect/downloads/create-gup-symantec-can-help-conserve-bandwidth-clients-remote-location
video on GUP
https://www-secure.symantec.com/connect/videos/group-update-providers-part-1
https://www-secure.symantec.com/connect/videos/group-update-providers-part-2
Regard
AJ
errata to my previous answer.
errata to my previous answer. the SEPM is in the 224 subnet mask, don`t know what i was thinking :(
the GUPs in question are in same subnet mask as 255.255.255.0
using VLANs because of different companies we supply solutions in same locations we use something like this
Company A has class 172.25.5.X ( using VLAN5 )
where x = 1 to 255
subnet mask 255.255.255.0
gateway - 172.25.5.254
Company B has class 172.25.6.X ( using VLAN6 )
where x = 1 to 255
subnet mask 255.255.255.0
gateway - 172.25.6.254
Company C has class 172.25.7.X ( using VLAN7 )
where x = 1 to 255
subnet mask 255.255.255.0
gateway - 172.25.7.254
now let`s take an example of Branch:
Branch 1 has offices for Companies A +B
so we have for company A : serverA1Branch1 (172.25.5.1), computerA1Branch1, computerA2Branch1,computerA3Branch1
so we have for company B: serverB1Branch1,computerB1Branch1 ...
the problem is rising on Branch X
company A : serverA1Branch67 ( 172.25.5.230 ) , computerA1Branch67 (172.25.5.231), computerA2Branch67 (172.25.5.232) , ...
all of these Branches are set to use a LU policy with Single GUP - each of the time is used the serverA1BranchX (where X = the local branch )
so computers in Branch1 from company A have set "single GUP policy" IP: serverA1Branch1
so computers in Branch2 from company A have set "single GUP policy" IP: serverA1Branch2
i understand the idea from the GUP tech data about "multiple GPU" settings where i can failover other GUPs in same VLAN, BUT i have on the server set up the policies with only "SINGLE GUP " and still sometimes computers in Branches 2 - 10 come to ServerA1Branch1 :(
maybe right now is more clear the image of the network and you can come with some ideas to prevent computers in Branches to jump to others servers instead of their own server. right now the only solution i had to limit the damage to the network was to unlink all Liveupdate Policies over the affected area where the 2 MB WAN connection was crippled by all the others WANs connecting to it
maybe i could lived with it if there was a way to limit the GUP transfer speed to the other computers, as i already throttled the speed from SEPM to GUPs at 512 KB and is perfect as it is.
so, any ideas ?
Would you like to reply?
Login or Register to post your comment.