[sims] Timeout on new server

[jayemm]

Ok, so we were one of the schools affected by SQL running at 99% due to our extensive use of panels on the homepage.

 

Capita's suggestion to fix this was to drastically reduce the number of panels used. My fix was to buy a dual hex core server and thankfully it has paid off and SIMS on curriculum runs faster than ever.

 

But we appear to be having problems with SIMS on Admin network, which is the network that the SIMS server is part of. Timeouts especially appear in Exams.

 

I have visited an affected machine and checked DNS resolutions, ping times, connection to server etc and cannot find anything wrong. I have checked and running the same things on curriculum or on the server itself work fine. I have also tried adding the connection timeout line to connect.ini but no change there either.

 

Anyone had a similar issue before and fixed it? Or any suggestions on what else to check.

[bossman]

Ok, so we were one of the schools affected by SQL running at 99% due to our extensive use of panels on the homepage.

 

Capita's suggestion to fix this was to drastically reduce the number of panels used. My fix was to buy a dual hex core server and thankfully it has paid off and SIMS on curriculum runs faster than ever.

 

But we appear to be having problems with SIMS on Admin network, which is the network that the SIMS server is part of. Timeouts especially appear in Exams.

 

I have visited an affected machine and checked DNS resolutions, ping times, connection to server etc and cannot find anything wrong. I have checked and running the same things on curriculum or on the server itself work fine. I have also tried adding the connection timeout line to connect.ini but no change there either.

 

Anyone had a similar issue before and fixed it? Or any suggestions on what else to check.

 

Have you looked at your AV? Could this be a reason why clients are timing out as the files on the server and also the client are being inspected by your AV each time they are accessed. Just a thought as you seem to have covered most other things. ;D

[jayemm]

I had thought about that, but the staff computers on Curriculum, and the computers on Admin, both have the same AV program. I will try switching the AV off on a test machine to rule it out.

 

Other suggestions still welcome, as I feel like I am running out of reasons!

[vikpaw]

If you're using the same server, and it runs fine on one client, but not on another, assuming they are identical, or as much as can be i'd be inclined to think it was the networking somehow. Something must be different, if you've tested networky things, then look for the differences, the users, their permissions, the policies that apply.

Can you move a curriculum machine to the admin network for testing, if you just let it pick up the ip range for that side..? Try the same domain account from the other side...?

[bossman]

What separates the 2 networks? Is it a switch (VLan) or a router?

 

Could it be the way the admin accesses the network through this switch or router which is different to the curriculum network?

 

Different protocols setup on the switch ports, packet sizes, frames etc?

 

How does the curriculum network access the Sims server what authentication does it use?

 

How is the Sims server configured for the admin network what is the difference between the clients?

 

Er! its still a little early and my brain is not warmed up yet will think of some other stuff

[jayemm]

The two networks are physically sepperate. Different switches, fibre and copper. Two physically seperate domains. The SIMS server is a member of the admin domain.

 

DC is a server 2003 machine, new sims is 2008 R2 with SQL2008 R2 SP1.

 

The switches are not managed just factory defaults. We have tried changing where things are plugged in and still no changes.

 

The admin machines map the drive and it uses domain groups to assign the share and security permissions.

 

Curriculum map the drive as local account on the SIMS server. To eliminate this I tried mapping the drive on an admin machine in the same way but that made no difference.

 

I wondered if there was an issue with the old server still being switched on, they have different instance and hostnames though, but this made no difference.

 

aaaaaaarrrgggggggh

[jayemm]

Hmm, ok. So we have found out what may be causing it. If I disable the curriculum NIC, admin works fine. If I disable the admin nic, curriculum works fine. When both NICs are enabled, the one that was enabled last has the timeout issues etc.

 

We never had this issue with 2003 so ideas welcomed.

[vikpaw]

Are they just separately configured IP ranges, or are you using VLANing controlled by the switch.

[dwhyte85]

Hmm, ok. So we have found out what may be causing it. If I disable the curriculum NIC, admin works fine. If I disable the admin nic, curriculum works fine. When both NICs are enabled, the one that was enabled last has the timeout issues etc.

 

We never had this issue with 2003 so ideas welcomed.

 

Sounds very silly but if you have teaming enabled - does it have a configurable mac address? Had a very similar issue with several supermicros... turned out that we had dup mac addresses from other teamed machines. Manually setting a mac address in the teaming options fixed it, try running arp -a and see if any duplicate macs.

[jayemm]

Hey, just thought I would give you an update. I followed some instructions about dual home setup from microsoft, but it all appeared to be ok except a static route on the 'secondary' network which was actually the one that was working. Decided to reboot server the switch the server is plugged into and the core switch and so far all appears well!?

[vikpaw]

Bizarre, so basically, the "Turn it off and back on again" strategy seems to have worked. Let's hope it lasts.